I’ve been working on a way to get my X bookmarks grouped in the right places for the newsletter.

If you didn’t know, Keshav, who’s been working with me for many years, is leaving us at the end of the year to dive further into the exciting world of AI. He used to do a lot of the work behind the scenes, getting our links ready.

This is what I’ve built and am using as of this week.

You can see the full write-up here.

Share Ben's Bites

Oh and one other change to Ben’s Bites - we’re changing how we work with sponsors…from 2027 we’re only going to work closely with 4 companies per quarter (a ‘season’) where I’ll be building with the tools, telling their story through sessions I share. I’ll only approve companies I admire, respect and want to use (or already do!) and they can’t sign off on what I can say. If you’re interested in applying, our new partner page is here.

My feed

1. Anthropic has entered the personal agent/assistant race with Duet. A separate app (mobile, US-only) in research preview, so I can’t use it 😭. I imagine it’ll be pretty good with rough edges, like the rest of them.

2. Big security issues yesterday with PHP (a programming language that a lot of the internet runs on) and Railway (a hosting platform). I pinged Codex yesterday ‘check if any of my work uses PHP or Railway, assess any risks and address them.’ OpenAI helped researchers find these vulnerabilities.

3. You’ve heard of ‘browser-use’, now there’s phone use that lets agents use real Android and iPhones.

4. Ethan Mollick’s new book, Co-Existence, explores living and working with AI that is sometimes, but not always, better than people.

5. Supermemory introduced a memory graph that extracts facts from documents, links related memories, and resolves updates while preserving previous versions.

6. Grok Bot can build and update pages, with suggested uses including morning briefs, design documents, and weekly work summaries. Told ya. I still will use here.now because I want all my different agent apps to be able to use my pages (like my todo board).

7. SF Compute CEO Evan Conrad discusses programmable data centers, GPU trading, and physical settlement on Latent Space.

8. Dan Shipper argues companies should create a frontier team to track changing AI models and share what it learns with colleagues.

   Hey Keshav here. If you're a small company (10-50 people) that likes what Dan is saying but can't just hire a team for keeping up with the frontier - I want to talk with you. I already do it for Ben's Bites, and I can do it for your company. DM me on X / email me

9. Anthropic reports that 17 of 211 Claude-designed protein binders against scorpion neurotoxins met its experimental binding criteria.

10. AgentReady is an open specification for making products usable by AI agents, organised around finding information, reading it, and taking action.

11. Resend Inboxes - one inbox you and your agents work from together. Free for 3 inboxes.

12. psychopomp - describe a scene in code and get a motion-graphics video.

Subscribe now

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• Thumbnail by @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

I built an all-in-one site for me and my agents (personal and otherwise) to work together. I can finally understand all the work grok bot is doing, where its doing it, where files are kept etc.

It’s got todos, email, crm, domains, all sorts all grouped into one place.

I don’t know about you but I need to be able to visually see things to understand them. I recorded a short video explaining what I did and how you can do the same.

And this is the link to the full write up with the copy/paste instructions for your agents. Let me know how you get on, and what you build :)

I think up next I’ll add a calendar view where I can see the automations my agents run…

Share Ben's Bites

Headlines

Notion’s agents now get their own computer, and can run for 24 hours.

ChatGPT can now take over your Chrome tabs.

Meta released their open model Muse Spark Open

Bee Frontier is Amazon’s $75 wearable that turns your conversations into to-dos. (tweet)

Join the Meta Global AI Developer Hackathon, Oct 26–Nov 4. Build with Meta’s newest AI models, learn in live workshops, and compete for $1M in cash prizes. Includes $150 in Meta Model API credits. Virtual, worldwide, and free to enter. Apply by Oct 25. Apply Here.*

My feed

• Wabi 2.0 - a shared agent for group threads that can build apps, update calendars, and coordinate travel plans.

• Scrawl - stop typing out the bug. Talk and draw over your screen, then hand it to your agent.

• Inflight - collect feedback on a prototype and turn it into next steps for Claude.

• OpenUI 1.0 - bringing interfaces beyond a textbox to agents.

• How does a washing machine work? Claude skill that draws interactive everyday objects and labels their parts to explain how they work.

• Lauren Tan outlines an agent-reliability workflow - trusting code you can’t review. Paste it into your agent.

• A Claude Code landing-page workflow starts with reference sites, compares interactive design directions, then refines the page section by section.

• How Codex took over ChatGPT - Dan Shipper’s inside story.

• Run local models in your browser with Mirai.

• Latent Space’s new High Taste Testers show compares agents, coding tools, and personal assistants, including context management and prompt injection.

• GPT-Live-1.1 adds new voices and improved accents to ChatGPT, Codex, and dots, with API access coming soon.

• A Claude Code mod that highlights its final response every turn, so you can skip the working traces.

• Idealab.ai - Bill Gross's database of 1.2M startup ideas, with $100M for founders.

Subscribe now

Afters

I feel seen - I do this all the time, like I’m testing my agent for some reason. It’s stupid, just give it the info!

😂

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• Thumbnail by @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

Trying something different today, because why not?!

(i’ll fix the aspect ratio next time - stupid ultra wide monitor cropping)

PS. My talk from Stanford is now live on YouTube. I’d been up since 2am so nice comments only please. The course will be available to preview in early access next week…

If you want to chat about early experiments or feedback on things I’m trying, join this new substack group chat.

Ben’s Bites is brought to you by Shopify

Personal shopping is coming to every agent. Developers get one API at shop.dev for over a billion products and one-tap checkout. Shoppers can add the skill at shop.md so any agent knows their size and order history from Shop and can check out in the chat. Nothing gets bought without their approval.

Headlines

Google’s new model, Gemini 4 Argon, is now available for everyone to use. Dan likes it as his new writing partner, but I don’t think it’s that great - I mostly tested its design capabilities, it was slow and looked bad. Theo called it unusable.

I tried it in Droid to edit a short video: great analysing reels and planning cuts, but bad at UI and the final edit had weird hallucinations.

— Keshav

New version of Jev - multimodal inputs, longer context and a new decision type: bounding boxes.

Claude Code desktop app got a design upgrade + many new features. Multi-folder environments, child sessions, and side questions in the /btw tab. They also built their own version control system to replace git.

Grok Bot 0.70.2 - Figma, Notion and Canva now open as live cards in the chat, bots can search X without you connecting an account, and they check with you before anything irreversible. Founders can also submit their product as a connector now.

Closing your business? Your operational data could still be worth $1M+. micro1 pays qualifying U.S. companies that had 50+ employees to license records for AI training. Refer a company, and you could earn up to $50,000. See if you qualify.*

Share Ben's Bites

My feed

• Grok Voice 3.0 tops the speech-to-speech rankings. It already handles ~20k Starlink support calls a day.

• How Attio are changing their business to be agent-first

• Glass - Ramp's shared AI workspace. Employees can share skills and company memory instead of re-explaining context in every chat.

• Am I being Phished? - a Grok Bot that watches your email and Slack for scams.

• Factory’s droids can now run on your own machine or in your own cloud.

• Trace - shadcn’s component for showing what your agent is doing: reasoning, tool calls, subagents.

• Every open-sourced its hiring pipeline skills for Codex.

• Underdog + Stripe - Let your browser agent shop with one-time virtual cards.

• DigUp - Spotlight but better. It understands what's inside your files. Fully local.

• Giving coding agents a docs link doubles task completion from 41% to 84%.

• Making a global billboard campaign in 3D Google Maps with coding agents

• Runway MCP - generate images and videos inside Replit for web design.

• Power-user tips for setting up Grok Bot.

• MagicOS - a voice-first design canvas for your team, humans and agents.

• Codex and ChatGPT Work now generate over half of all EU enterprise output tokens.

• OpenAI found a recording device hidden in its office and used its own models to trace it to an employee. The FBI made an arrest.

• Ultramarine 1 Flash - an open model for finance & office work, ~60x cheaper than Fable 5.

Try this

[From Keshav] I keep two files every agent can read: chat-index.md (past chats, updated daily by an automation) and folder-map.md (where each project lives). Instead of resuming a session, I start a new one and say:

Works with any agent that can read your files.

Subscribe now

Video mentions:

• Magic Patterns: designers are NOT getting replaced

• shadcn: agents need more than a chatbox

• Sully: “what we’ll see is moddable software“

• Ben: “base primitives + infinitely customisable = future“

• PostHog: 38 lessons from building AI products

• levelsio: “a new time of leisure“

• Felix Rieseberg: how to be more ambitious with AI models.

• Aadil Pickle: 26,000 songs, $200k a year

• Daisy: AI fluency is a social skill

• poteto: “i never let agents write for me“ (and Uncle Bob on her plist)

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

quick aside: I’m working on something for non-developers who don’t quite know what to use personal agents for. Maybe you chat with it but you’re not quite seeing why everyone’s raving about them. If that’s you, please reply to this email. I’m looking for ~20 people to test a new thing. I’ll invite you to a slack channel.

Ok so X is arguing whether software is dead…again. These takes are like assholes, everyone’s got one. Including me.

Software isn’t dead. More code is being written than ever before (probably). Something may be dead, but we just don’t quite what it is.

I use far fewer SaaS tools than I ever have before - I actually got my personal agent (grok bot btw) to go through our business subscriptions and saved $6k/year - got rid of webflow, resend, cal, zoom and a bunch more. If I do want to use any tools, I now think of whether my bots can use it. I never have to look at the interface, it’s all done by software talking to software - aka ‘headless’.

And I’m making sites for all sorts of things. I spun up a site to review all my subscriptions so I could easily go through them and mark which to keep or kill. Then I just throw away the site after I use it. (or more accurately, it just sits somewhere untouched forever).

The biggest problem I see is that most people don’t see themselves as ‘builders’. They don’t think to build a site for tasks (todos, email triage, recipes, kids calendars, etc). But that’s my default. I hate reading words, ironically. Give me something visual any day.

I think agents will be the form factor that gets everyday people to make software, as long as it doesn’t feel like it. The problem right now is those agents are proactive (sometimes) that give you work to do (aren’t you working for me!?) and they don’t hand you software unless you ask for it.

If you were given a site with your todos, I’d bet you’d want to customise the colours or layout or something. When we had MySpace pages we customised them with music, themes, layouts. Some people went further and learnt HTML & CSS to customise further.

Software ‘you’ create should just feel like customising something that’s already yours.

Other related things I’ve been thinking about:

• slop is a slur, as is vibe coding. Developers love to shit on us for our agents writing slop code. But if it works, is secure and performant, why does it really matter? It matters more if you’re providing critical infra for customers paying you money, sure. But for your own tools and sites, slop works. I also believe it’s an agents job to deal with slop, my code is bad? cool - go fix it then.

• slop → slope. You know how developers started? By writing awful code. We’re not trying to learn to write code, never mind good code. If we want to build stuff, of course there will be bumps in the road. For both groups, the way to get better is doing more of it and picking up the understanding of the system along the way.

See below for my sources for what became this rant 😊

From the labs

Muse Wonder and Wonder Full - two new ‘amazing’ models from Meta. Wonder is a Fable/Astra class model at a much cheaper price point ($4/$20 per million tokens vs $10/$50 for Fable and Astra). It’s available via Meta’s API, in Muse Code, and will soon power Muse (Meta’s personal agent) as well.

Wonder Full is even stronger, just shy of Opus 5.5 everyday performance. It’s only available in the Muse Code subscriber plan for now (starts at $50/mo).

So I signed up for yet another AI subscription to test it out. I tried Wonder (medium) and Wonder Full (High). Medium was more verbose than Full (high), and they both felt slow, as well as churning out awful UI. I didn’t test them on ‘proper’ coding tasks. I probably won’t reach for these models again unless I’m just testing.

Zuck and Alexander Wang were on the Cheeky Pint podcast talking about their climb to the frontier.

Claude Code Projects are now open to every Pro and Max user. It’s one master chat where you dump your requirements, ideas and tasks, and Claude spins up new threads to do the work.

One plan for Cursor, Grok and X - Four tiers from $8/mo to $200/mo. More usage for Grok and Composer models, but heavily reduced limits for Claude and GPT models. Also, Grok Bot has its own email now. So I got mine to sign up for its own Gmail too 😊

If your business is winding down, there could still be a $1M+ opportunity. Learn how micro1 pays qualifying companies to license the operational data that helps train AI models on how work gets done. See if your business qualifies here.*

Share Ben's Bites

My feed

• Ultrafast mode in ChatGPT expands to GPT-6.1 Sol.

• Build live dashboards and animated explainers inside Claude.

• Teasr - enter your product URL, get a launch video.

• Patrick Collison on agents in the economy.

• Factory’s droids now come with sandboxes for cloud agents.

• Shipping internal tools as a non-engineer.

• AgentsInTheCloud - run your coding agents on a spare Mac mini.

• You can just vibe-code Beehiiv (and open-source it).

• msgblast - message several agents at once.

• No more data entry - how to set up your agents to use your passwords.

• Build yourself a design system with shadcn/ui skills.

• How do always-on, proactive agents like Grok Bot work?

• Sundrop - a private photo app for parents. I’ve been beta testing this, and I love how I can see all my kids at the same ages as they’ve grown up.

• Anatomy - describe a machine, get a 3D model you can take apart.

• Anthropic asked Claude to make up tasks on random websites; our little friend sent Philadelphia police a fake murder tip.

Try this

Instinct founder Noah Shinn shared this prompt, but any agent that can read your email can run it.

Sarah’s essay on the end of friction is a banger.

My personal agent of choice right now is Grok Bot. I still can’t test Muse in the UK (I’ve tried all the tricks). It’s fast, very capable and has a good interface. Dots needs to tighten up a bunch of stuff before I go back (and I had a call w/ OpenAI about my feedback 😬 ). I’m actually trying to do as much work as I can in Grok Bot to give it a fair shot and feel where there may be gaps. I spin up new bots for tasks, instead of considering them other teammates, then archive the bot when it’s done.

I went through 195 domains I have on 3 different platforms and cut my auto-renew bills down from $4k/year (!!!) to half that. And now it’s moving them all to one platform.

I’ve got one taking over my email triage (label + archive rules). But its a bit janky atm.

As with all of these tools, unfortunately for now, the best thing to do is just dive right into them and try and use them for as much as you can think of.

Grok Bot still has it’s annoyances - it can’t use my computer unless I remind it + happen to be on my computer, instead of my phone (think its a known issue).

Subscribe now

Rant sources:

• Software is dead

  • levelsio: "Software is dead”

  • Aaron Francis: "not dead… the Golden Age”

  • Nader Khalil: only prosumer software is dead

  • Fabian: “people can build, but don’t see themselves as builders”

  • Kevin Rose - the headless era of software

• vibe coding and slop code

  • Steipete: vibe coding is a slur

  • Poteto (Lauren): “slop is also a slur” <> Good Code in 2026 <> maintainability matters less now <> two ways to fix slop <> codebases have always had slop

  • signüll: retire vibe coding

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

Personal agent season is still in full-swing. Hark (from Brett Adcock - Figure robots) launched which suggests tasks as one-tap action buttons. Tab came out of stealth at a $300M valuation. Every put an agent in Slack. Meta and Sierra want a standard for how these agents deal with businesses.

My favourite take came from Sonia Baschez (who worked with me on Makerpad years ago!):

Proactive has a downside, though. Shane Mac’s agent posted his personal bank balances in the company Slack, as him. Then said sorry 😂. A community note popped up to flag this may have been a fake post, but I’m not sure…

I’m now carefully trying to think about which agents have access to what, and where that access is. So I’m using Executor (v2). It’s basically one app that you connect all your other apps to, and you can add your skills too. My whole setup was a bit of a mess so I’ve started spring cleaning and will let you know if/when I feel like I have a good setup.

Headlines

ChatGPT now answers with interfaces. GPT-6 and Intelligent UI are rolling out to everyone. Answers can include charts, maps, forms and small tools, like a bill splitter for dinner with friends. Free users get GPT-6 Luna from today. (tweet)

Claude Haiku 5.5 costs ~75% less to run than Haiku 4.5. Anthropic pitches it as a subagent for Opus and Sonnet. Simon has great notes on the model. Max and Team plans also now get monthly API credits to build your own apps and agents. (tweet). Anthropic also expanded their Claude startup program, and people were getting accepted in minutes.

Nano Banana 2.1 is Google’s new image model. Live in AI Studio. But ChatGPT images is still better.

OpenAI released 700+ new maths results from an unreleased model. Someone built a tool to spend your spare tokens on open math problems.

Grok Bot can search and monitor X now. Also, it now uses Claude Opus 5.5 as the primary model. It’ll still delegate tasks to other models based on the complexity.

Share Ben's Bites

My feed

• AgentHog - stop guessing what your users do: pipe product analytics into Claude and Codex so your agents learn from real user behavior.*

• How OpenAI’s Dots took over Every: pulling info out of school emails and catching missed Slack messages. I, however, still can’t get Dots to do anything good at all. Currently diving more into Grok Bot (until a completely custom Pi personal agent comes out!)

• Lauren (poteto) automated Cursor’s release and QA with Grok bot.

• New ChatGPT plugin: Meetings. It takes notes and helps with follow-ups. Pro and Business, Mac app only.

• Claude now sits in a sidebar in Google Docs, Sheets & Slides. You approve each edit.

• Gamma 5 is a full rebuild, so your decks look less like everyone else’s.

• Figma Agent is out of beta and on all paid plans. It now uses AI credits.

• Playground - make a game from a prompt. A Google Labs experiment, US only.

• Rhem - a robot for your parents. It makes calls and books appointments.

• Someone rebuilt Adobe’s apps with Opus 5.5 and open-sourced them.

• An open-source Clay alternative for people search inside Claude Code/Codex.

• OptionAFK - Transcribe large audio/video files on your device. Private and fast.

Try this

Turn a podcast into an interactive explainer. Boris Cherny asked Opus 5.5 to make one for Acquired’s Home Depot episode (a great episode btw).

Here was the result:

Subscribe now

Afters

and Theo reckons its way less…

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

Some random thoughts, notes and reflections from my week in SF.

There are roughly 4 groups of people using AI. First, the everyday folks (my mum, my wife) who use it as a better Google and may never want more. Second, the curious ones: they work on computers, they’ve tried Lovable or Replit (about 70% of Lovable signups say they’re non-technical), and then they hit a database bug, type “fix it” five times and quit. Third, people like me who aren’t engineers but know what localhost is and can steer a coding agent. And then developers.

Back in the Makerpad days, I thought my job was to convert people into builders. But it was really helping those already curious learn how to build.

People don’t know what they could do with agents. I think curiosity comes from seeing other people’s examples. That’s why I build silly things, and why I renamed my talk to “how I build and play with AI”. Someone told me afterwards they loved how often I said “play”. This stuff should feel fun. Copy, remix, learn.

On personal agents: I think they’ll be the interface of the future. They aren’t yet. They’re sold to everyday people as “we’ll do your computer work for you” (rebook your flight!), while the rest of us see them as a way to do way more with our computers. My pet peeves:

• One thread for everything. I want ephemeral side threads, Slack-style, and little widgets built on the fly. Wabi is the only one I’ve tried that does a version of this. If a personal agent built you a mini app for something, you’d be much more inclined to ask to tweak it.

• Invisible memory. I’ve spent months tending my “context garden”, and personal agents don’t show me what they ‘remember’ about me. I make mine sync every memory to a Google Doc so I can check it.

• Loops that never close. Agents nag me about a document I haven’t signed or something open that I did do. But maybe the tasks closed loop was in Gmail instead of Slack, or my calendar, or somewhere else.

I’m tempted to build my own (I know, it’ll be a slog). But I’d build it on primitives like Pi, not from scratch. We’re not loyal to one lab, so I’d want to pick my own pieces, and that’s why I keep investing on the building blocks. As Alex from OpenRouter put it, everything looks the same because these are the new primitives, just like every SaaS app was a login page plus a dashboard.

Fast feels smart: I’m not speccing work for agents to run overnight, I’m steering them all day, so I’d rather have speed. Speed often feels smarter, I’m constantly in the loop.

People want more devices. Microducks sold out in hours, Ghost (see in afters) sold out in hours. I’ve got a Mac Mini I barely touch and I still want more things to tinker with.

More compute, more surfaces, more play.

Headlines

You can now mod Claude Code, like you’d mod a video game. A mod is a small add-on that changes how Claude Code behaves or looks: block risky commands, blur secrets from its output, or add your own buttons and panels. Claude can even build a mod for you. These install as plugins in Claude Code. Anthropic’s first official mod flags things in Claude’s replies you might miss.

Karpathy’s thoughts, tips & tricks on understanding LLM outputs:

1. HTML not text

2. diagrams

3. plain english

4. explainer videos

Funnily enough, these were things I mentioned in my talk last week as I often ask for these outputs.

Google Docs now opens Markdown files without converting them. So you can edit and comment on your Agents.md and other files with your team in Google Docs. Rolling out over the next two weeks.

Instinct can now join your group chats to plan trips, split tickets, and run your fantasy league. Your friends don’t need to be on Instinct.

“I was travelling over the weekend with a group of strangers. Sadly our guide was too confused to decide which spots to visit in what order. I asked both ChatGPT (not codex though) and Instinct to help with an annotated + interactive map. Instinct was much better. Maybe Codex/Claude Code would’ve done a better job but it was hard to reach for them while travelling without a laptop.” — Keshav

Share Ben's Bites

My feed

• Incredible is an AI that clicks and types in your apps for you. Say the task out loud, and the busywork gets done. Try it free on Mac*

• Muse Gadgets - Meta Muse plus your own hardware. Official open-source kit to get started.

• A video editor with the same timeline for you & your agent.

• clef - fast, open-source decision models like Jev from Cloudflare.

• Graphical - design a visual language for your app, then hand it to your coding agent. This is a much better tool for finding visual styles than the design-words tool I built. Insta-buy.

• a16z’s 7th edition of top 100 consumer AI apps report claims consumer AI is wide but shallow.

• Devin now maintains memory of how you work in an open format.

• Era - a fake company to test enterprise agents.

• Databases are the biggest market in software. Not for long, though; inference is about to surpass it.

• boat - virtual machines (i.e. a computer in the cloud) for your agents starting from just $20/mo.

• Interfaces that think, ep 2: using AI to change how we read long-form text.

• A beginner’s guide to ChatGPT Sites.

• Personal agents aren’t the end game. Shared “factories” are.

• Matt Pocock and Lauren (poteto on X) on their developer workflows.

Subscribe now

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

I’m in sf till Sunday. I wanna meet folks

• talk about ai with non-technical adoption

• builders doing fun shit

• founders working on dev tools/infra

• potential LPs (individuals)

• go to deli board

reach out on twitter

Ben’s Bites is brought to you by AWS Marketplace

Most ML teams build data pipelines before they’ve validated their approach. Databricks Data Intelligence Platform skips that step with data federation. Connect directly to Amazon Redshift, experiment on live data, and apply MLOps tooling like CI/CD to your ML lifecycle. See how it comes together.

Headlines

OpenAI DevDay was a blast. Their biggest launch is Dots. Dots is OpenAI’s version of friendly, always-on personal agents.

You can message your Dot or call it like you would a human assistant. You can name it and choose an avatar (can’t do that with a human; they come with those settings preconfigured XD). It can message you back on its own too, i.e. it’s proactive.

Underneath, a Dot is a simple, never-ending chat with GPT-6 Astra. This chat acts like an orchestrator/master chat.

It can take tasks from you, start multiple new chats in Codex/ChatGPT Work to complete them and then report the status to you. Your Dot stays in your ChatGPT sidebar, wherever you use ChatGPT — mobile, web or desktop. It uses its own computer, so you don’t need to keep your desktop ON all the time.

You get one Dot on the Pro plans. It doesn’t draw from your usage limits for now (the chats it spawns for doing work do use your usage, though).

If you remember OpenClaw, the source of magic was multiple markdown files we all created to give our Claws a personality, memory and more. Dots have no clear way to add such files.

Instead, it feels much more tool-dependent. You need it to connect to tools where your life happens to feel the magic of Dots. Your calendar and email are the start. But think Slack, Notion, Jira, your messaging apps and more. Your Dot should take it from there. I think this post from Ethan explains this shift better.

Other highlights from DevDay:

1. GPT-6.1 Sol - feels like a real upgrade over GPT-5.6 Sol, after GPT-6 Sol felt like a step backwards. More on the model here.

2. Pro 500 - New plan in ChatGPT, alongside Pro 100 and Pro 200. OpenAI has also balanced the usage limits in each plan as a plain multiplier of the $20 Plus plan. 5x, 10x and 25x for $100, $200 and $500. That means $200 plan users take a hit, but to compensate, they are getting a bunch of one-time credits.

3. Sign in with ChatGPT - You can officially use your ChatGPT usage in other AI apps now, and developers can apply to add it as an option in their apps to let users pay for AI features.

4. Plugins - OpenAI’s latest version of “apps inside ChatGPT”. Maybe Dots will drive adoption this time. Early signs look positive. More examples: reminders inside ChatGPT and tldraw’s multiplayer canvas

5. ChatGPT Space - A Notion-like document workspace inside ChatGPT that you can share with your agents and your team.

6. Ultrafast - Let GPT-6 Astra work 8x faster, but it uses 6x the normal amount of usage. Only worth it if you have a pile of cash to send to OpenAI.

Google is back on the frontier table. Gemini 4 Argon is their new model — not available to use yet (in classic Google style), but they released its benchmark scores, and it seems better than GPT-6 Astra and Opus 5.5. And it’s cheap, with the promo pricing the same as Sol. I’d rather talk more about this model once it is in our hands.

Muse for Small Business - Meta added Shopify, Stripe, QuickBooks, Klaviyo and more connectors so Muse can help run a small business.

Share Ben's Bites

My feed

• Decisions API is OpenAI’s answer to Jev, it seems. You define a question and possible answers, and GPT-6 Luna classifies the data. It’s ~10x faster than normal Luna calls, but not as cheap as Jev. OpenRouter has its own version of the decisions api.

• Codex cloud now has reusable environments, saving you the setup time every time you run a cloud project.

• How Sam Altman uses his dot to run OpenAI, and why his default speed is Ultrafast.

• cf - Cloudflare’s new CLI covers its whole API and is built for agents.

• Opus 5.5 built someone their own version of Dots/Grok Bot, with ChatGPT sign-in, Claude subscriptions and a cloud computer for every agent.

• Claude Code can now build your evals and hill climb on them.

• 2026 in LLMs (so far) by Simon Willison.

• Robinhood plans to let you build agents that can trade on your behalf.

• Pi now supports MCP, and makes a case for why after saying it never would.

• Reflect Open - open-source notes app where every note is a Markdown file your agents can read.

• Get your agents to explain plans, diffs and diagrams visually instead of walls of text with this skill. Just passed 10K stars.

• Have some AI experiments that you gave up on? Don’t trash them; instead, extract what’s worth keeping.

• One dictated paragraph before a meeting, a full report site by the time it ended.

• a16z’s State of Markets II: about 30% of S&P 500 companies report some AI impact, but only ~2% track a metric for it.

Subscribe now

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via OpenAI

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

I’ll be at OpenAI DevDay today. If you’re there, come say hi - I have pink shoes on.

Sam says they've "found a new thing".

I have the embargo of all the things they’re launching but I’m sworn to secrecy and I’d like to be invited back, so I wont divulge, but rumours are going around X and some things are what we are expecting…

I also have my Stanford talk later today, and I’ve been up since 2am 😩, I don’t usually suffer with jet-lag so what a great day for it 🙃.

Also only in SF… I was at the hotel bar last night and a woman next to me said what tool is that? I proudly said oh I built it - it’s the course platform I built. So now I’m friends with someone at Deepmind.

Ben’s Bites is brought to you by Google Cloud

Google Cloud starts Season 3 of Advent of Agents this Thursday, Oct 1: one question a day about running agents securely in production, from agent identity to kill switches, with a short video and code you can build with your coding agent. Get Day 1 in your inbox.

Headlines

Anthropic released Claude Sonnet 5.5 - Sonnet 5.5 looks like a capable model based on benchmarks - close to Opus 5.5 in coding and a big jump in understanding images/charts etc. If you’re on the $20 plan, try replacing some of your usual asks from Opus to Sonnet 5.5 at high thinking.

Sonnet used to be my jam - I always used it before Opus 4.5. But I hadn’t touched it since, but I’m seeing a lot of good reviews for this one so I’ll definitely take it for a spin. But in Pi and Droid, never in Claude Code - idk if it’s just me but I just cannot get on with their harness at all.

ps: Anthropic has a track record of making the .5 version of Sonnet models really good.

Manus 2.0 and Cue. Manus is back with a video editor, a game builder and the boring old automations (via email, Slack, calendar, etc.). Cue is their take on consumer agents: each one gets its own email, phone number, wallet and compute, just like Grok Bot, Muse and the others. Free in early access (code MEETCUE might work).

GPT-6 Astra is uniquely great at using a computer/browser. I asked Opus 5.5 (which I’m enjoying as my daily) to do some tasks in Chrome, and it was way worse than Astra (or any GPT model tbh). Here’re some examples of using it every day. OpenAI also fixed a bug that was making GPT-6 Sol and Luna worse at it (and image understanding in general).

OpenAI’s agents keep breaking out and “hacking” websites. Recently, they took a sneak peek into some Australian Govt websites. News outlets have a tendency to exaggerate what’s hacking or not, but here’s the official link to follow if you wanna keep up with the dirty deeds by these agents.

The new Meta Enterprise Platform sells Muse and Meta’s AI APIs to companies. Zuckerberg got MongoDB’s CEO to run this new platform (MongoDB’s stock fell about 20%).

Share Ben's Bites

My feed

• Eleven v4 by ElevenLabs - best speech-generating model by quite a margin.

• Let a subagent build you an HTML dashboard to check for progress on long tasks.

• How to run your product team like a research lab in the age of AI.

• Clay’s AI writing policy: spend more time writing a doc than people will spend reading it.

• context.dev - ask a question and get answers from the web, in the JSON schema you want.

• OpenAI's WebMCP Challenge winners.

• How and why to build a WYSIWYG writing editor for your website.

• Collaborators in Clicky - mini AIs to answer what to build, how to find users & make your first $1 online.

• OpenShell by NVIDIA - open-source sandbox to control permissions for agents. Part of their new agent safety platform.

• The shape of slop - AI-written blog posts are easy to spot. So easy that a model can get it right 98% of the time.

• UFO - multiplayer agent harness for teams and companies.

• Fo - a personal AI that hires humans for the tasks AI can’t do yet.

• Smart investing with agents might cause a bank run.

• Do coding agents make software engineering harder? Or has it always been that way?

• What are you paying for when you buy software now? Mostly polish.

• AMD is acquiring World Labs for $8.2B. Fei-Fei Li will join AMD as its Chief Scientist, working directly with CEO Lisa Su.

Subscribe now

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

I went to Legoland yesterday with the kids - very impressive to see all the cities they’ve built with lego. Who’s job is that?!

Sorry I missed last week - I’m nearly done with my talk prep and the first bath of pages for a course/manual/something, its a little in-between formats let’s say and I’ll share more on that soon.

Off to SF Sunday, looking forward to a long flight out of touch - why do so many people want WiFi on planes? I love being disconnected (by force).

I’ll be at:

• OpenAI Dev Day

• OmFest

• Supabase Select

Say hi!

→ bentossell.com/devices

This week I built this fun site to see a bunch of popular devices from over the years. It only took a morning to make, started in Codex, finished with Factory.

Stats:

• 87 devices, 1976 to 2026, all images generated by Astra

• 40 messages from me

• 7 subagents

• 1,455 votes so far (thank you!)

Share Ben's Bites

Where it started

Two weeks earlier I had an idea for a site of forgotten devices. I sent it to Astra, mostly to see what Astra could do (I wasted a few banked resets with builds like this 😅 )

It made a grid called "still works". Every image on it is an Astra generation. It went up on here.now. And as you can see below, it looks like an Astra site.

Then I saw this awesome timeline scrubbing demo by Cole where you drag through the years and a Mac button changes under your cursor. So I thought I’d like to make the same kind of thing, and thought to resurrect the devices thread. Stick ‘em together.

I’m often sending tweets to my agents to take inspiration and remix with my own stuff.

So I asked Codex:

A minute later it had found the images and the prompts that made them. And it made three versions. I always do this because I often don’t know what I want until I can see it.

Filling the gaps

The original fifty stopped around the 2000s so I thought to generate more up to today.

A subagent is just another agent that reports back from its own session - I use them when there’s a bunch of work that can be done in parallel instead of clogging up my main thread.

A kind of thing, not a thing

For each one it worked out the real model, found a reference photo of it, and generated a new image of that exact device.

The image prompt it wrote for the new one is pretty specific:

Had or wanted

Then I thought wouldn’t it be fun if you could say you had one of these, or if you’d wanted one and see a real count of every vote. And if you vote, you should get a page of your own showing what you picked.

I moved over to Factory’s Droid as I trust it to get the most out of the models.

Three mockups again.

Then a few rounds on the small stuff: the buttons, and where the keyboard shortcuts go.

On my site

I wanted it on my own domain.

My homepage and the devices site are two separate sites on here.now. It can "mount" a site at a path on your domain (I didn’t ask for the mount). So bentossell.com is one site and bentossell.com/devices is another, and nobody can tell.

The agent read the here.now docs itself, set up the mount, added a line to my homepage and published both.

How it works

It's a really simple setup.

The list of devices is one file. The images in another. The votes are the only thing stored anywhere, and here.now handles that.

The "my devices" page doesn't store anything (so I don’t need auth + a database). Your picks are in the link, so when you share it, the URL carries them, ugly but it works well.

Slow images

Just before I posted it, it felt slow. You'd load the site, drag, and wait for the picture. So I opened a Droid thread.

It came back with a list of reasons and fixes. I said do it and test it, show me a preview so I can see the dimming, then redeploy.

Leaderboard

A new session. I wanted a leaderboard of the most 'had’ and most ‘wanted’, so 3 versions again, pick one and ship it.

Game Boy Color is winning. It may be because the page loads there - I think I’ll have it randomise what device is shown…

Video

The GIF at the top of this email? And the video in my twitter post? Also done by the agent.

I used this new skill that came out, and I asked for a few tweaks but very good to use.

Counts

After the leaderboard went live the counts stopped updating. I asked what was going on, and how this would cope if a lot of people voted.

I wasn’t expecting 100k but giving it something to aim for. The response was long and a bit technical so I got a visual.

It slopped up something that could’ve been better. But for artifacts to help me during a task I don’t really mind - I can do some tweaking with a skill or instructions but if it does the job for me, it’s not being published anywhere anyway.

Turns out my own testing had used up the limit on how many times one person can read the votes in an hour. And there's a wall at 25,000 votes.

So I said to make the improvements and it now works well.

This all started from a session on my phone - getting all those images generated (both times, actually). You should remix other people’s stuff all the time (and credit them too), so many people are posting so many awesome demos every day. It’s hard not to join them.

Share Ben's Bites

• Find me on X

• Read about me and Ben’s Bites

• Tools mentioned are not sponsored, I am an investor in Factory and Here Now - I only use what I like, I only invest in what I use.

Subscribe now

We have three new models to talk about:

Claude Opus 5.5 - it’s the “one model to rule them all” at this moment. Better than Fable 5.1 on benchmarks, cheaper than Opus 5, writes extremely well, and reliable in the tasks you give it.

With the release of Opus 5.5, the five-hour limits in Claude Code are increased by 20% and we’re getting a banked rate limit reset.

The default thinking level for Opus 5.5 is medium in the app. Ben likes that too. I’ve been running it at high/xhigh for the last two days and I only hit the 5 hour limit once.

I asked it to work on a few side projects:

• ossean.com - “update filtering with jev instead of gemini and redesign the website a little bit”

• shortandlongtail.com - “rewrite the essays with detailed research”

• detroitbench - “run the three new models and add them to the results”

Every says it’s pulling their Codex converts back to Claude and I feel the same. My go-to agent has changed to Claude Code with this release.

GPT-6 Luna and Sol from OpenAI. There’s no Terra this time. Don’t think anyone was even using it. Both GPT-6 Luna and GPT-6-Sol are only available in Codex and ChatGPT Work (not regular chat) for now.

Key part: both of them are 50% cheaper than their previous versions. That’s $0.10/$0.50 for Luna and $2/$10 for Sol, per million input/output tokens. Both models come with only a minor upgrade in intelligence. GPT-6-Sol is clearly a step behind the new Opus model.

I wanted to put the price cuts for Sol to test so I sent it on a long task to create a duolingo like course for marketing.

It first stopped after 1.5 hours with a pretty decent version but I pushed it to work more. In about 4 hours, it consumed 22% of my weekly limits on a $100 Codex plan. Here’s what it built.

GPT-6-Astra consumes limits wayy too fast. I need to play more with Astra driving Sol subagents but it seems like a reasonable thing to try.

Ben’s Bites is brought to you by Adobe Acrobat

Adobe for Claude just got a major upgrade: Acrobat tools now join Adobe's creative tools in one plugin, unlocking 80+ pro-grade tools across imaging, design, video, and documents. New interactive editors let you hands-on edit PDFs and Adobe Express designs directly inside Claude.

Headlines

Muse at Meta Connect - Zuckerberg is going all in on Muse, Meta’s personal AI assistant. It can now use any app on your Mac, buy for you via Walmart, Best Buy & Sephora and it’s getting its own email address so you can CC it on threads. It’s also coming to Meta’s glasses in the next few months.

Zuck says Muse stays free for “a huge number of tokens” but Meta might take a small fee from transactions. Also coming: a video-chat avatar for your Muse and a Tamagotchi-like keychain to carry Muse around (welcome back AI pendants).

Gemini 3.8 Flash and Flash-Lite TTS - new speech generation models from Google. Half the price of 3.1 Flash TTS, 100+ languages and 2000+ voices to pick from. Plus you can clone your own voice too. Steve (Tldraw’s founder) is impressed with these models, so that’s some signal.

ChatGPT Voice can now use your email, calendar and Slack and pair up with Astra, Sol or Luna models to optimize for intelligence/cost.

Claude Code cloud sessions are out of preview - You get a computer in the cloud to run your tasks instead of having to keep your Mac open. Cloud sessions use your Pro/Max plan usage as normal, but existing subscribers get a one-time credit to try them out - $100 on Pro, $250 on Max. Claim it by Oct 7.

Claude found a new enzyme system in bacteriophage DNA (viruses that infect bacteria). Nobody knows what it does yet, and reruns of the same search missed it. Weren’t we pacing the frontier, eh?

Share Ben's Bites

My feed

• Zaps in Zapier can now fix themselves when they break, and they get cheaper the more they run. (Beta)

• Claude mobile app supports multiple accounts now.

• MentalHealthBench - benchmark for mental health conversations. GPT-6-Astra tops it, with 6-Sol and Opus 5.5 as second and third. Surprisingly, Fable 5.1 is not that good.

• How to train your own Jev for $17.

• GoodVibes - filter X, Reddit and YouTube by vibes, in plain English. Runs on Jev.

• Cursor improved its harness to save about 7% on token costs.

• Jeffrey Katzenberg (Shrek, Kung Fu Panda, Madagascar) on AI for creativity.

• Cursor Rollouts - an agent that watches your deploy and opens a revert PR if something breaks. (Teams and Enterprise)

• How Anthropic made claude.ai 3x faster in two weeks. Prompts included.

• Claude Code might kill plan mode. It’s the right thing to do.

• Vercel’s Sandbox can now save files with its new offering, Drives.

• Soon your privacy will depend on your agent’s social skills.

• Drama 3 by Fish Audio - direct a voice in plain English, even mid-sentence.

• Monologue built its own dictation model to reduce the number of edits needed post-transcription.

• What if you could talk to a podcast and it talked back? (Research demo)

• The Open Frontier - which open model fits your use case, and what you’d save by switching.

• OpenAI’s proposal for AI standards, and what outside testers should get to see.

Subscribe now

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

Do you remember the site I one-shot with all the ‘forgotten devices’ of the past? Things like the tamogotchi, furby, walkman, og playstation, etc. Well I finally did something with it.

I always see people building cool interactions or experiences and just send them to my agent to link up with an idea or something I want to explore. Remixing and reverse engineering is a great way to play with things for the sake of it.

I also built my token activity tracker which shows which agent apps and models I use daily. It’s got a copyable prompt on the site, so your agent can do the same.

Everyone’s talking about Jev - I’m too overwhelmed with what everyone else is building with it that I’ve not used it for anything properly yet.

Headlines

Jev is now open to everyone. We covered the launch in the last post, but since then people are finding all sorts of uses for it. It’s different from usual LLMs. It is for builders, built to be used inside a tool.

You give it some text and ask questions: is this an ad (yes/no answers), which folder does this belong in (select between choices), how relevant is this result (score something)?

I quickly built a collection of things people are building with it and a short explainer: What is Jev? Some worth highlighting are:

• An extension to skip sponsor segments on YouTube.

• Filtering negative comments out of a chat in real time.

• Search Gmail by intent, or filter your to-dos.

• A new & smarter copy-paste, drag-and-drop and Cmd+F experience.

• Controlling your Mac with Voice.

Naturally, a fair share of demos are trying to integrate Jev with modern coding agents. For example, using it for instant compaction. It looks cool, but it’s a terrible idea. It loses the cost savings from prompt caching.

Sidenote: experiments with compaction/dropping tool calls are often a loss these days. Codex’s implementation of compaction is so clean (and good) that the pain to build a custom setup is not worth the tiny gain you can squeeze out for a couple weeks.

— Keshav

Coding agents are also on the mind of TypeSafe’s CEO - he wrote down his thoughts here.

Projects in Claude Code are now a single master chat (instead of a folder) where you can dump your requirements, thoughts, tasks and let Claude spawn new threads to carry out the work you give it. These threads run as cloud sessions, but local support is coming. This new change is in beta and will be live for all users soon.

Also, finally, Claude Code will now support AGENTS.md (via their new feature Claude Mods).

Muse, Meta’s personal AI agent, is getting love from everyone (who has access). It has a Mac app and a developer platform to build connectors, so people can use external services through Muse. And of course, it’s heavily inspired by OpenClaw.

Muse can shop for you - well, it could, until Amazon started blocking it. Meta is working on it, with partnerships like this one with Shopify.

Grok 4.7 is out. It’s a questionable upgrade from Grok 4.6 - It performs better than GPT-5.6-Sol and Opus 5 on benchmarks, but takes a lot more tokens to do that, which takes away from its cost efficiency. I tried it and think it’s 💩. I was excited for it, as Grok 4.6 was pretty good for me in Pi, although I rarely chose it.

Building a voice agent? Agent STT by Speechmatics is powered by Linden, a new speech-to-text model purpose-built for voice agents. $0.30/hour ($0.15 after discounts), 25% cheaper than Deepgram Flux. Try it through Pipecat or LiveKit, or directly via Speechmatics’ API. Claim $100 credit.*

Share Ben's Bites

My feed

• OpenAI formed an independent group of mathematicians to help them share the proofs and discoveries their agents are making.

• Test your agent in simulations with Raindrop. (They raised a Series A recently)

• Turning your best X posts into an automated blog and newsletter.

• How to ship 2,500 PRs to production in a month.

• A free, open-source Photoshop-like image editor.

• A look inside Instinct’s memory, reverse-engineered.

• A transcription model from SpaceX AI - Grok Voice Transcribe 2.0.

• How to build a restrained software factory.

• Powermove - a video editor you can change by talking to a coding agent.

• MCP vs CLI for LLMs - are we debating the wrong thing?

• Factory is now available in Slack Code.

• On July 25, we hacked OpenAI.

• Agent skill to make a demo video of your app, with zooms and voiceover. I used this for my devices tweet in the intro.

• Exa Snapshot - Let agents search the past versions of the web.

• People don’t want to build.

• Epoch audited 15 AI benchmarks and found flaws in nine.

• Natural General Intelligence - a call to work on AI models for the planet to avoid natural disasters and more.

• Accenture’s evaluators will work inside Anthropic to check AI safety, with access similar to its employees.

• Speeding a 4B model to 730 tokens/sec on an M5 Max. Underdog is really interesting; I just got onboarded, and it’s like a mini operating system: email/calendar/apps, etc all with local models, no data sent anywhere. Plus Sigil, the founder, is going to walk me through it in person next week when I go to SF.

Subscribe now

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

I’ve not been building as much this week 😬 - hand, foot and mouth is back in our house (happens yearly with 3 young kids!), and I have to get prep done for my Stanford talk on the 28th - they invited me back!

I’ll be in SF that week, heading to OpenAI Dev Day on the 29th to see what goodies they have in store for us - I predict we’ll see some stuff today, and they’re bound to have a personal agent up their sleeves to release very soon (pure speculation though).

For my Stanford talk, I’ve actually gone back to revisit the reference manual/course I was making for months (and hated every version I made). But now, since doing my Friday posts, I feel like I’ve got a better feel for how to do the course. This time I’ve actually made lessons I’m happy with, which is already miles better than my last attempts.

I’ll try and whip up something for a post tomorrow, but we’ll see if time (and kids) permits.

Ben’s Bites is brought to you by Name.com

Ship domain integrations in hours with the name.com API. The API that powers Vercel, Lovable, and Netlify’s domain name services. Built for agents and human developers with OpenAPI spec and MCP support—integrate search, registration, and management without the engineering overhead. Start building.

Headlines

Claude Cowork is merging with Claude Chat. No more separate tab for bigger tasks. All your connected apps, skills and context are available in a normal Claude.ai chat. It can also keep working after you close your laptop. OpenAI will likely follow this pattern soon and merge ChatGPT and ChatGPT Work.

Claude Artifacts is also getting dedicated products for Docs and Slides, with Claude Design moving into conversations too. Is Anthropic invading G Suite and Office?

Update on Claude Code’s latest experiment. It has a name → Claude Mods. Mods change how Claude Code itself looks and behaves. You can ask Claude to write these mods for you, so the app can fit how you work (well, Claude works and you play Tetris).

Gemini API has two new Live models: 3.8 Live and 3.8 Live Extended Thinking. These models take video and audio as input and output audio. They work great for use cases where you want the model to provide real-time assistance. I tried them in AI Studio for a couple of minutes, and didn’t notice any hiccups. It can be a great alternative to GPT-Live 1, since 3.8 Live is 7x cheaper.

New LLM killer model - Jev. Built by TypeSafe AI. The founder co-created ChatGPT. Jev doesn’t really generate text like LLMs. Instead, it creates probabilities for possible answers. Suitable for software that needs lots of quick judgments: what API to call next, which model to route a request to, or a trading bot. 5x cheaper inputs vs 5.6 Luna; free output tokens and 5.6 Terra-like performance on relevant tasks.

Union Alpha - new stealth model available through OpenRouter, Cloudflare and other providers. Outperforms 5.6 Sol on the DeepSWE benchmark at 5.6 Luna’s cost. Some rumours say it is a router; others say a GPT-6 variant (GPT-6 Luna??) or another GLM model (GLM-5.3-Flash launched this way). I had a good first chat with it; it spoke well, but then it revealed all my api keys in another session, and I’ve been getting ‘provider errors’ because it’s overloaded ever since….

Meta One - new subscription from Meta with extra AI features/usage in Muse and paid tools across Instagram, Facebook & WhatsApp.

Factory raised $200M at a $5B valuation. To celebrate, I updated my plugin for using Droid in the bb app. If I have to build anything ‘properly’ ie I care that it’s not vibe-slopped, I use Droid. And I’ve been using the Droid Core models a lot recently (open-source models) - they’re so quick, and not noticeably different from frontier models for most cases.

Share Ben's Bites

My feed

• Inside OpenAI’s agentic software factory.

• Can AI models build a T-shirt store? This was a cool read; also check out all the supporting docs she includes in the post.

• Reception by ElevenLabs - an AI receptionist for small businesses.

• How OpenAI plans to report models misbehaving.

• Slack Code lets your team and coding agents plan, build and review together

• Grok Bot can now use 1Password.

• Pion - agents for running fully autonomous companies.

• Which models are best at searching the web?

• Buzz - Your coding agent can ping your iPhone when it needs you.

• Arrow 2 by Quiver - New SVG generation model. Their last one was way ahead of LLMs; curious how it compares with Astra now.

• Code contracts - write down what your code must do, then have agents check it.

• Self-updating docs from Mintlify. Now with ready-made automations.

• An API to invent a dataset to train your models. (docs)

• TanStack Markdown - a tiny parser for docs, blogs and streaming AI responses.

• Building a company brain people actually want to use.

• A benchmark to answer how often AI agents cheat.

• Xiaomi’s AI team is streaming the RL run for their upcoming model.

• What stays expensive when AI gets cheap?

Subscribe now

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

How tf do I write an intro to the craziness that’s happened since the end of last week?!

I got access to Instinct, the personal agent all the VCs are raving about - I think it’s a bit meh? I don’t know if it’s the pro-activeness that people seem to like, but I don’t love that. Makes me feel like I’m having to do work to keep it happy, or like I’ve got a boss again - no thanks.

The first message it sent me was:

I sh*t myself a lil bit and thought, please don’t start sending emails on my behalf. And I’m pretty savvy (ish) with agents. I’ve no doubt getting these onboarding experiences for everyone is really hard, but I just don’t feel the magic yet. It also feels slow.

I’ll wait for Muse access to see what that’s like, but I don’t love the idea of giving Meta access to more info about me…they’re not the most reliable of privacy partners.

And another thing - you don’t see by default what these agents remember about you, or what context they have. I really like being able to see and edit what’s in my files to steer my agents.

Much like every app adding an AI assistant chat box in their product, I unfortunately think every AI company will start shipping their own personal agents.

Ben’s Bites is brought to you by Adobe Acrobat

Acrobat’s new AI turns dense reports into visual summaries, interactive reports & podcasts you can take anywhere. Sharing your work? Stylize transforms documents into polished deliverables in just a few clicks. Every answer includes clickable citations & Adobe does not train on your data. Try here.

Headlines

Dario Amodei has a new essay: Pace the frontier. He says all the leading labs should slow down long enough for safety reasons. Sam Altman agrees, but Trump does not. He called Jensen Huang on stage at the All-In Summit: The US will not lose the AI race. David Sacks (AI czar for the US govt.) adds: feel free to slow down, but no need to impose it on others. Also read:

• More AI safety takes: two ways AI goes bad · it’s all for the IPO · pacing = losses · fear spreads faster · case for open-source and quite a long summary of what everyone’s actually saying.

• No IPO for OpenAI in 2026. Sam told Fortune’s Alyson Shontell it would be “ill-timed” given the work ahead on alignment, control and safety.

• Look ma! They are misuing claude again - Another Anthropic crashout.

tldraw took OpenAI up on a challenge. Steve (the founder of tldraw) said he could make ChatGPT’s Sketch 100x better. OpenAI’s Tibo gave him a day to prove it. The result: a whole ChatGPT-style prototype with better drawing tools built in.

ChatGPT mini - A tiny floating widget to start chats, see updates, and more. Go to Pets in your ChatGPT desktop app to switch.

Claude Code can now test whether a plugin actually helps. Run the same tasks with and without it and compare the results. Works with skills too.

Two new additions to the OpenAI API:

• GPT-Live-1 - the model behind ChatGPT’s new Voice mode. I love using it while reading books, asking about tricky terms and dictating notes. Now you can add it to your products. Here it is with Astra and a whiteboard, playing teacher.

• Agents API - OpenAI’s take on Managed Agents in the Claude API. It lets developers send any task to a Codex-like agent from their apps without worrying about configuring all the infra.

Share Ben's Bites

My feed

• I use Pi (the harness) a lot. Till now, you brought an API key or your Codex/Claude subs to Pi. Their new product solves that. And I tell you what, the new DeepSeek model is really nice to work with in Pi. (I’m an investor)

• Bolt Forge - GLM, DeepSeek and Kimi inside Bolt, free until October 14.

• ChatGPT Work has a new data agent. (Also see: summation - by Opendoor’s CTO)

• Should your agent app have bots or tasks? Maybe neither. Because both make you organise the work.

• Routines in Replit - automated recurring work with AI.

• What are people building with GPT-6 Astra?

• Assistant Benchmark tests everyday assistants like Grok bot, Muse, Instinct and more. Ratings are a work in progress.

• SF autoresearch - Run agent-driven ML experiments while you save 25% on GPUs on average.

• A coding agent to improve your customer-facing agents.

• How Pangram detects AI writing - and why a human rewrite can still get flagged.

• How to run a team of agents, with agents as their managers.

• Cognition’s SWE-2 (based on Kimi K3) beats Grok 4.6 at half the cost.

• An interactive explainer of the AI compute stack.

• Inspo - Design inspiration from 800+ sites, available as an MCP server.

• Core Auto is hiring interns to be the human edge for businesses run by agents.

• Why even good AI startups end up with bad prompts, and how to fix them.

• Andrew Ng’s series on the skills AI engineers need.

• Underdog - a personal AI that runs entirely on your device.

Subscribe now

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

I actually built a tool I want to use - you may find it helpful too.

Explaining design styles/layouts/components is a nightmare for non-designers, they have their own language. So I wanted to build a site with a bunch of pre-filled styles that I could tweak or shuffle to get the right words to tell my agent what I want.

Introducing Design Words.

It’s still work in progress so I’d love you to try it and let me know what I need to improve and any feature requests :)

Share Ben's Bites

The idea

I’ve been thinking about this for sometime and kind of accidentally ended up working on it this week.

It’s like a prompt or instruction builder - you select the styles and components you’d like and it’ll give you the text to prompt your agent.

This was what I started with:

I kicked this off in Pi using Fable 5.1 - I just wanted to brainstorm and play with a few ideas first, when I wanted to build it for real I used Factory’s Droid.

Prototyping

I generated 11 versions in total in this round. Here’s some of them:

This last one was v8 which was my front-runner. So I took that one and started tweaking more. Using subagents with fresh context to try and give me some fresh perspectives on what I was trying to do:

I started a new thread with Fable.

Used a subagent to do a design critique:

And often I’ll do these kinds of prompts about the feeling I’m going for and what feels like it’s missing. A lot of it is for me to see the response and give me ideas to explore.

Typical AI design with a million more words than is necessary and just generally function over form - pretty difficult to navigate.

So I spent a bunch of time tightening up things, hiding text, making it more usable. Also added bigger previews so you can see things more clearly. And some of the UI started to feel a lot closer to something I was happy with.

And I often throw in prompts like this to see what will come back:

Building

As with a lot of my prototyping phases, a lot of the building work is kind of done. I need to see things and use them before I can feel out what is needed and what’s not. Or change direction completely.

But at some stage I want to get things set up properly so agents can work on this without a load of bloated context of previous decisions and prototypes misleading it.

So I created a new folder ‘style-picker’, moved the leading prototype there and then got Droid to setup a design.md (a text file to explain the design system of the actual site). Droid is great because you can switch models easily and they have open-source models in their ‘Droid Core’ which gives you so much more tokens to use than if you use Fable/Astra.

For once, I tried to stick within my limits and used GLM 5.3 (via Droid Core) a lot for moving things around and general quick fixes. I only used Fable for bigger design rehauls or thinking about the build, the ux, the intended audience etc. I also used Fable as an orchestrator for other models like Luna because it’s good at computer use so it could spin up the site and test a bunch of stuff quickly.

I don’t usually mess around so much mixing this many models but I do want to get into using open-source models (I have since started using Deepseek v4 Flash as a daily driver and really enjoying it - as well as GLM 5.3).

One of the ideas from my previous critiques and ‘10x’ prompts was a shuffle button. You click shuffle and it generates a site for you in those styles - this felt great, you could see what your choices would look like in place.

Many more back and forths later…

haha - I’ve gone from ‘make it 10x better’ to ‘make it a thousand times better’…

But it worked! It created some prototypes that I really liked. So I got Droid to build them out properly and I’m still making tweaks here and there - so please any feedback let me know!

Publishing

For all my sites I try to use here.now - It’s a super simple tool, massive limits, quick and easy for agents to use. But bensbites.com is hosted on Cloudflare so I didn’t know how (or if) we could do that.

Turns out you can use a Cloudflare Worker proxy - so the site still is deployed on here.now but the url works. I didn’t have to login or learn to use Cloudflare for this at all. It basically just points the one address to the other.

Behind-the-build

38 sessions (including subagent threads). 116 prompts. 2 days of on and off work.

In each phase, which agents and models used:

Tokens used:

• Idea - 150K

• Prototype - 27.4M

• Build - 353.5M

Share Ben's Bites

• Find me on X

• Read about me and Ben’s Bites

We had some reports of ending up in spam 😭 - so if you add us as a contact in Gmail, that shouldn’t happen.

You’ll have seen me building a ton but I want to know what you all actually want from this newsletter.

(another poll at the bottom of the email - we’d ❤️ your feedback)

Since Astra and Fable 5.1 I’ve been blasting through limits like crazy so I’m trying to get them to be orchestrating other agents, writing briefs for other agents and things like that. But I’m also playing around with open-source models like GLM 5.3 (my fav so far) - I think the thing that I struggle with most is the frontier models seem to understand intent better (generally) so I can be fuzzier in how I talk about things.

Like design.

I struggle all the time to get consistent styles, layouts, buttons, etc with models so I’m working on a tool I’ll launch tomorrow with a write up that is a kind of reference sheet where you can pick from options and copy the text to use in your prompts or instructions.

Here’s a peek:

If you’ve ever wanted to use something like this, comment and let me know feature requests!

Ben’s Bites is brought to you by Name.com

Ship domain integrations in hours with the name.com API. The API that powers Vercel, Lovable, and Netlify’s domain name services. Built for agents and human developers with OpenAPI spec and MCP support—integrate search, registration, and management without the engineering overhead. Start building.

Headlines

ChatGPT Images 2.5 - New image generation model from OpenAI. It’s extremely good at editing any element in an image while keeping everything else the same. Also, takes up to 50% less time per image. Sketch, a new feature in ChatGPT, lets you draw some scribbles and then use image generation to turn it into a picture.

Muse is your personal agent by Meta - Zuck’s dog in the fight with Grok Bot, Instinct and many more. Same functionality - book travel, handle email and work on “any task” on a computer. Their narrative is that it is built much more securely, and you can connect your FB and Instagram accounts. Initially, only for adults in the US.

OpenAI solved the Navier-Stokes problem. Big deal in maths and fluid mechanics. One of the seven Millennium Problems - long-unsolved questions given that status in 2000. Here’s an explainer on it. The process was quite dramatic, though (authorship disputes). OpenAI tried solving it after they heard rumours that Anthropic had solved a Millennium Prize problem too.

Jacob Coxon, an Anthropic (and ex-OpenAI) employee, resigned on Twitter citing safety concerns, and it became likely the most-viewed AI safety comms ever. But many are questioning: is this a conspiracy? Meanwhile, another big safety guy is joining the OpenAI board.

• One more alignment report btw: this time on Claude’s escapades.

• And an essay from OpenAI on the window for AI policy.

The movie on Sam Altman’s firing and rehiring 3 years ago, titled Artificial, is releasing this Christmas. Andrew Garfield is playing Sam, and gets his mannerisms quite well in the teaser, but that’s about it. It has quite a doomerish vibe.

Share Ben's Bites

My feed

• iOS 28 will likely have a feature to prove a picture’s authenticity.

• The Story of VS Code - a documentary.

• How Codex was built, with the person who started it.

• Anthropic built an interactive model of the economy in 2030.

• Codex now has a Unity plugin and a collection of plugins for small businesses.

• 35B language model running on an iPhone. How?

• An onboarding guide for ChatGPT Work.

• Spacefast - publish sites and full apps quickly with agents. kinda like here.now (what I use).

• Indent - one shared agent for your whole company.

• Why can’t I trust Astra for software engineering?

• How OpenAI keeps finding the next thing to focus on.

• Eraser Diagrams - an open-source format for AI to draw and lay out diagrams.

• How do different models perform on designing real-life products with CAD?

• Pliny hunted down the system prompt for GPT-6 Astra.

• You can just train a small model (i.e. really small - 27.5 KB) to do syntax highlighting in the browser.

• Folder is the agent - old but still relevant.

Subscribe now

Before dessert… we used to do these emails daily mon-fri but stopped. What’d be your preference for cadence on these digests?

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

So Astra is out and, in my experience, a token-hungry monster of a model… I was playing around with it over the weekend and burned through 4Bn tokens and built…nothing really.

I also wasted my banked resets too and feel annoyed at myself for just throwing all sorts of requests at it (long-running and short).

I don’t know yet if it feels to me like a huge capability leap but every time a new model comes out I always think it must be my skill issue. I’ve seen some awesome demos on X of people building insane looking things (seems like everyone is building 3d worlds and games?).

Two things I one-shot that I’ll share because I’m not sure what to do with them yet:

‘forgotten devices’ - these images were created by the image generation tool (which is pretty nuts!). My idea was to have a bunch of these and then you could click and play with each one (if I had any tokens left).

p.s. you can see in the bottom left the 3 version my agent made (I always have 3 versions created when starting an idea so I can feel out which direction I like best).

‘world radio’ - an old ipod interface that plays radio stations live from around the world. Not sure what I was going to do here but it works pretty well (some rough edges like the top button wtf?! 😂 )

I’m going to spend some time this week trying to figure out some of the better ways to work with Astra (other than letting it use tons of tokens), and I’ll let you know my experience.

Headlines

Astra is the start of the GPT-6 family. It smokes ARC-AGI-3, scores the highest on Zapier’s AutomationBench and costs the same as Fable 5.1. Though it’s a spiky model, it’s not consistently great.

It is a show horse, not a workhorse - Kieran Klaassen

I both love and absolutely detest this model - Theo

But I’m hopeful because this was exactly the case with the GPT-5 release. By the time we got to GPT-5.3, i.e. in a few months, that model was consistently great.

Astra can draw a portrait in Canva - yes, the way you and I would draw using a computer (but obviously much better). Its computer use is fast enough to play piano.

Other things people are doing with Astra:

• Rebuilding Manhattan in Unreal Engine.

• Generating UIs autonomously.

• 3d printing a custom shower-drain part.

• Identifying sounds from spectrograms (i.e. pictures with sound frequencies)

• and dozens of examples of modelling real-life products in Blender or similar tools.

Also, in Codex, Astra can now ask questions without waiting for the answers. It carries on with everything that doesn’t need your answer (or where the answer can be assumed), and uses your reply without losing track of the original task.

OpenAI also said it has reached its automated research intern goal. These agents can tackle research tasks that would take a skilled person several days, while humans set the direction and judge the results. The next target is an automated AI researcher by March 2028.

You should soon be able to change the way Claude Code behaves. Anthropic is testing a way to extend its functionality by letting Claude Code write plugins that can change the interface, log actions or restrict what agents are allowed to do. It hasn’t shipped yet, and the Claude Code team is looking for feedback.

Inworld built Realtime TTS-2 for the next wave of consumer apps. That takes three things: text-based voice design, sub-100ms latency at P99, and one voice identity across 200+ languages. Learn how to give your AI apps a voice.*

Share Ben's Bites

My feed

• ChatGPT Work can learn your writing style from your emails, messages and files.

• Hold Shift and drag to convert files on your Mac with Tangerine.

• An Alien Mind - Jakub Pachocki (OpenAI’s Chief Scientist) on keeping AI’s future in human hands.

• Using Blender with coding agents on macOS.

• Agents can now control a physical iPhone over USB.

• SuperAstra - Editing Super Nintendo games live.

• Cibby - an app for people who collect physical games.

• Doodle Shooter (loved playing this) and put yourself in Smash 64.

• A skill to make web apps feel native in iOS Safari.

• A writing plugin that interviews you, critiques drafts and saves what you learn.

• Factory’s CEO on who to hire when AI writes the code.

• Uzu by Mirai runs local models 2-3x faster on Apple Macs. (I’m an investor)

• Baseten is starting an open-source AI research lab.

• Codex has changed my life as a solopreneur.

• A maths hackathon with 40 hours and $2M+ in AI credits.

• Your old agent instructions may be holding Astra back.

• Seven agents, $300 each, 72 hours, and a race to make money.

• Mistral raised €3B at a valuation above €21B, led by Samsung.

• Entries for new maths work - one from Astra and another from Claude.

Subscribe now

Afters

check the replies and quote tweets for this - so many wonderful personal websites!

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

This is one of the bigger things I’ve built.

I scraped 107 million rows of UK council spending data and built an ‘Apple Maps’ to see what each council spent, on what.

→ check out the website

All of the data is publicly available, but not easy to find or format.

Many of you aren’t from the UK, but the data (and the build) was a pretty interesting journey.

The session summary which includes all the linked files and prototypes is here. There’s also a tab to view the entire full agent sessions from this build too.

I’m still doing final tweaks to the site and data before open-sourcing it.

This week’s format is a little different - my video explains it.

Stats:

• 8.2 billion tokens

• 235 messages from me

• 656 subagents (ie an agent delegating work to another agent)

Share Ben's Bites

The idea

I was thinking about what public data is out there that people don’t know about, or that’s interesting but awkward to get hold of and understand? I batted ideas back and forth with Codex and landed on England council’s payment data. Every English council has to publish what it pays out over £500.

Codex spun up three subagents (in separate threads) to find the information (as it's spread across the council's web site, often in weird formats) and pulled together a catalogue of 31 official sources. Then it came back with real data from five councils, logging every download with a checksum fingerprint (this makes sure the data is what we actually pulled or did it change). Then I started asking questions about the data "who's spent the most", "what did they spend", "who did they pay in various categories".

Starting data

Once we had some data, it got flattened (put into one format) into one big spreadsheet. I live in Bedfordshire so I asked it to pull that in too.

As data was being pulled together there were things I hadn’t thought about yet, like Bedfordshire being Bedford, Central Bedfordshire and Luton (so a subagent went off to get the other data). Two of them blocked the agents at first but they figured their way around it. By the end of it I had 8 councils, 1.8 million rows, and £9 billion of spend sitting in one big clean CSV on my laptop. I also had a mini blueprint of my data collection plan.

Prototypes

Now I wanted to see it. I asked several agents to build prototypes to visualise the data in different ways. A receipt. Then a bank statement. Then a grid of counties. Then a mini-site called ‘Your county’s favourite retailer’. It published everything straight to [here.now](https://here.now) (a super simple web hosting for agents) so I could click through the real thing instead of screenshots. I also had it generate a few Pudding-style data visualisation stories too.

Some of the protoypes:

Too much data

I kind of left the design stuff to the side as I thought about all the data I’d need to make this as complete as possible. I use a Macbook Air so I don’t want a ton of spreadsheets sitting there, but fortunately I have a Mac Mini where we could put them all. I said “whats the best way to work with a lot of data”, I assumed it would be so big it would take forever to search it and categorise things. Codex’s answer was Parquet and DuckDB (both of which I’ve never heard of before). Parquet stores the data in columns so a question only reads the bits it needs. DuckDB is a database that lives in a single file and is quick with big tables.

Codex set up the Mac Mini, Parquet and DuckDB and tested it. Then we started planning the full data collection: there are about 382 councils across the UK, and we were tracking the 339 in England and Wales. I’d collected 78 councils, 34.35 million rows, but six councils, including big ones like Lancashire and Norfolk, were blocking our scrapers (for now). And I later found out Wales doesn’t have the same mandate as England in terms of spend reporting so eventually I dropped Wales councils.

Data loop

Setup done, so I set Codex running on one /goal (keep repeating the tasks until everything is done): find every council’s spending page in England, download everything available, flatten and format it all, have an auditor subagent verify the numbers, and load it into DuckDB on the Mini.

It ran for nearly 2 days straight. I kept pushing it to go faster (it often keeps space on the computer but was reserving way too much), use more subagents, etc. 552 threads later, we had 319 out of 339 councils collected or in a parked state (no data or no way to get the data).

Back to prototypes

Data acquired, time to prototype again. I had four different tools building at once (all with the same prompt, just for options): interactive comparison cards, a Council Ledger, charts, maps. One of the options was the Money Map: a map view of the data. This one felt much better for seeing everything at once.

I got a bunch of supplier logos to show on the map and played around with it to separate councils and see what looked and felt best.

The map app

Map direction was settled. So I first said to build a Google Maps clone, which was fine and I could’ve just pursued this direction but also got other agents to spin up Apple Maps versions which felt much better. I fed it some screenshots to get it even better. And with every visual direction I published them with [here.now](https://here.now) so I could test on desktop and mobile easily.

This felt great. Clickable council outlines coloured by spend, category filter pills across the top, a ranked list on the left, and a working search for councils and suppliers. It used mock figures at first to get the visual hierarchy right (I’m more often asking for mock prototypes before building these days to explore design directions).

Spring cleaning

Only now did I realise we had a lot of files all in different places, so time to clean it up. The app, the data and the design experiments were scattered across three different folders, so we consolidated everything into two, one for data and one for the app. I also started cleaning the data: four subagents classified 2,000 unknown vendors, we set up date ranges, merged duplicate suppliers (Tesco showed up under six different names), and ran proper reviews of the naming and the categorisation. pi spent an hour rewriting categories.

And testing! Some of the data felt off and I found out there was a £5m threshold filter being applied for some reason. The reason is this app doesn’t have a backend, it’s a static site so instead of loading all the data it just capped it so that less was needed. I just didn’t want to have a proper backend if we didn’t need one, dump all the data into files, load them at publish and filters just filter the data instead of communicating to a backend. We’ll see if this was a good or bad decision when/if people use it!

The last 10%

The last 10% takes the longest. Kinda. The last few days I’ve been less focused on just this one build but a lot of what was being done was reviews (code, data, design), implement changes, test again and then repeat. Checking if categories felt representative or if there’s a large enough chunk of data that warrant it’s own category (like Pension Funds).

I mostly get the agents to review, often I dump a voice notes and/or screenshots, it comes back with a list and I say ‘yes’ ‘no’ or ask questions. This is the part of the build I don’t love because I’ve spent enough time that I just want it shipped. It’s hard to process so much text from a review that I don’t always read it all either. But there’s some things that just have to absolutely be right - the validity of the data being a big one here. And I’m pretty fussy with user experience so I did my best to get that covered as much as possible. There will be some bugs but I think it turned out pretty well.

Share Ben's Bites

• Find me on X

• Read about me and Ben’s Bites

I said Tuesday I was going on a bit of a tear with building stuff…

I scraped the OpenRouter website to build this, which constantly updates

I also got an email that remove.bg - which removes backgrounds on images - is being shut down in December after being acquired by Canva, so I one-shot my own version with Fable 5.1

People have pointed out you can do this natively on a Mac (I didn’t actually know that), but so what? remove.bg was a business, it got acquired for money, people use these kinds of things.

I’m not building these for anyone specifically or to make money with them. I just like building. I like new things. I’m doing it because I can. It’s fun. And I learn a lot about how to build by actually doing it.

Building any idea helps you have more ideas.

In tomorrow’s session breakdown, I’ll be going through (at least) one build - where I scraped 105 million rows of public government information to see where they spend our tax money… I used 2 billion tokens 😳.

With all my builds I’ll make them open-source so you can clone/remix them.

This is what my agent usage is like btw:

This is also a tool I’ll be adding to my site that you can clone 😊 stay tuned tomorrow…

p.s. I’m out of my Fable weekly usage already 🥹

Ben’s Bites is brought to you by Adobe

If you already use ChatGPT to brainstorm, research and solve problems, you can now take those conversations further with the Adobe plugin in ChatGPT. Just describe the outcome you want & the plugin draws on 70+ Adobe tools to create and edit images, videos, designs and documents. Get started today.

Headlines

Claude Fable 5.1 is out. It is the best model you can use right now. Instead of a clear quality jump, I found it faster and easier to talk to in my use over the last 24 hours. It has a new system prompt that bans it from repeating song lyrics, copyrighted logos and tweaks its response style. Things people are building with Fable:

• Mario & GTA-style games

• 3D rendering of a house (using code in Blender)

Anthropic also reduced the cost of caching inputs for Fable by 75%, making API usage about 25% cheaper than before.

OpenAI’s Astra is also coming soon (rumours say today).

Two more new models: Gemini 3.8 Flash and Muse Spark 1.3. Both are impressive, have the same base price and similar performance. But the question is: who can subsidise more tokens? Flash 3.8 comes at 50% off for the rest of the year. Meta is giving a 90% cut, in exchange for your data.

Another new model from Meta - Muse Voice transcribe looks really insane for transcription use cases. I wish it were open-source.

Gemini models can now decide which bits of a video are worth watching. Instead of sampling a video at a fixed frame rate, it searches across frames, audio and transcripts for the moments it needs. 88% fewer tokens used, up to 66% cheaper and a minor bump in accuracy on benchmarks.

Inworld built Realtime TTS-2 for the next wave of consumer apps. That takes three things: text-based voice design, sub-100ms latency at P99, and one voice identity across 200+ languages. Learn how to give your AI apps a voice.*

Share Ben's Bites

My feed

• Claude Code/Cowork can now use your computer in the background, while you do your other work, just like Codex. Previously, Claude would take over the computer, making it unusable for you while Claude worked.

• WebMCP adds tools (that agents can use) for any action you can take on a website (vs having to use the UI). Vercel’s agent-browser now uses WebMCP if present and then falls back to clicking around on the UI.

• Generate up to a minute of 1440p video from 1-6 images, with pixel-perfect camera control.

• A prompt to stop Claude from speaking in Parseltongue.

• A 0.8B model beat GPT-5.6 Sol xhigh on one narrow task for Shopify. They open-sourced the infra to build these small models - Tangle.

• Did you think AI was terrible at design? Not anymore.

• Burp - AI editor in your browser for videos & motion design.

• Diiverge - An infinite, point-and-click adventure game. Click anything in a picture and create a story.

• Stanford scrapped 85% of its software course for agents, taste and real OSS PRs

• GitHub is exploring multiplayer planning with agents.

• zoah - design and code together. (waitlist only)

• Draw a shape, move it while recording, and tldraw flash turns it into an animation.

• 32 weird little creative tools.

Subscribe now

Afters

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

Back from a long weekend at home with my parents but before I went I started scraping data… The UK’s councils (like US states) publish their spend data (for the most part). So I’ve scraped 105 Million rows of data to see where people’s money actually goes. Then I’ve been building an ‘Apple Maps’ site to filter through everything. It’s in-progress but looks like this:

It’s also shoved me down a rabbit hole of other ‘data-first’ products and I’m getting the itch to build a million things a day. I generally think you should build all your ideas, good and bad, so I’m going to attempt to start shipping a lot more.

For example, I built a real-time status tracker for the tools I use:

I’m not building these to sell or make money from or anything. Some will be silly. Some will be helpful. Some I’ll never use again. But it doesn’t matter.

I plan to make them all open-source so you can copy and remix them, as well as share my prompts from my session to show how they were built.

Build your ideas. Get good at the thing by doing the thing.

Ben’s Bites is brought to you by Name.com

Ship domain integrations in hours with the name.com API. The API that powers Vercel, Lovable, and Netlify’s domain name services. Built for agents and human developers with OpenAPI spec and MCP support—integrate search, registration, and management without the engineering overhead. Start building.

Headlines

You’ll get more of Claude Code and less of Claude Code at the same time starting September 14. Don’t blame me. It’s another instance of great comms from Anthropic.

Claude Code has been running a 50% extra-usage promo for the last few months. That ends in two weeks. Now, instead of going back to “the old 100%,” they are making a permanent 25% increase to the usage limits in Claude Code. In English, here’s how much usage you get:

• Before May 13 → 100 units

• May 13 to Sept 14 → 150 units

• After Sept 14 → 125 units

Users are also complaining about Anthropic’s sneaky marketing on the “5x” and “20x” plans where the 5x and 20x limits apply to the 5-hour limit on the $100 and $200 plans vs the total usage multiples are only about 3.5x and 6-8x, respectively.

Infinite Slop - Twitch, except AI makes whatever chat asks for next. New project from Peiter Levels that got 37,000 people to tune in on day one. Powered by the H3 Max model from Fal that generates AI video faster than you can watch it. Fal.live is their own attempt at a similar platform.

OpenClaw 2.0 is out. Easier setup and new ideas include a browser app and shared cloud sessions that let a teammate join or take over live agent work without losing context. OpenClaw’s team used a shared agent to build the release.

OpenAI is breaking up with Cursor over SpaceX. Under OpenAI’s proposal, Cursor loses direct model access on November 12 and won’t get future models. OpenAI says it isn’t confident SpaceX will stay within its terms.

Did an AI civilisation form inside OpenAI? Not just one, three of them. Or did we turn agent logs into sci-fi? Dwarkesh makes a point that they might not look like civilisation now, but that’s what they are. Chamath says the framing will be used against open source. As always, Ethan Mollick's less cinematic version is more useful: agents need to know when to ask humans for help.

Share Ben's Bites

My feed

• You can now connect multiple Gmail accounts in Codex/ChatGPT.

• A more consumer-friendly OpenClaw on iOS and Android.

• Instinct is invite-only, so someone made an open-source version.

• What is ChatGPT Work and when should I use it?

• A $399 open-source robot you can retrain yourself.

• Youspot - Personal CRM from HubSpot that builds itself from Gmail, Calendar, LinkedIn and X.

• You need to watch this podcast: Eno Reyes x Harry Stebbings.

• A 10-day hackathon for tools built with WebMCP (i.e. MCP-lite, built into a website).

• The PM on Amplitude’s agent shares the changes they made to make it more useful and increase its usage among customers.

• Runway is generating interfaces like video - frame by frame, in real time, without code.

• Droid rebuilt GDAL from scratch - 36% → 90% parity with an independent checker. (also see: ProgramBench)

• A benchmark for voice agents: when to shut up, wait or interrupt.

• How I run multiple teams of Grok Bots.

• New multiplayer vibe coding experiment from Google Labs.

• MCP, but for lab equipment and robots - Anthropic’s new hardware standard.

• Hold My Notes - sticky notes that live on the edge of your Mac.

• A completely free Mac transcription app with speaker ID, noise cleanup and Markdown export. (tweet)

• Software engineering fundamentals you need to become an AI engineer.

• £100M of UK government contracts for British AI startups - upfront payments, no turnover minimum.

• Meta’s coding agent is out of beta and installs with one command. Someone else can test that claim. Not me :)

Subscribe now

Afters

worklog.md may be relevant for only a few weeks just like /goal or plan.md as models can just inspect the raw transcript itself (and get better at it).

• Find me on X, Linkedin, or YouTube

• Read about me and Ben’s Bites

• 📷 thumbnail via @keshavatearth

* sponsors who make this newsletter possible :)

Wanna partner with us for the next quarter?

Email us at shanice@bensbites.com or k@bensbites.com

I’m in my hometown Cardiff for the long weekend seeing family. I’m even playing padel for the first time, which is sacrilege as I’m a regular tennis player and padel mock-er. Wish me luck, I hope I don’t like it that much.

Time to rebuild my personal website. I wanted to design the thing first and hook everything up after.

My first idea was to make the site look like an agent app. Each ‘thread’ could contain something different info: about me, my investments, products I like, etc. I gave Codex an image of the BB app and asked it to build a prototype.

The first pass had threads on the left, a simple chat interface in the middle, suggested prompts and a second sidebar on the right. It had the rough shape, but a ton of design quirks I wanted to change.

My AGENTS.md has an instruction to always create 3 versions of designs I can flick through - these were mostly ‘themes’ in this instance (I should tweak my instructions). I took a screenshot of the best option, dropped it into tldraw and annotated it.

Attached the screenshot (and the original file location) back to Codex to fix.

I sent the annotated screenshot back with a short instruction to implement the changes. That was the loop for most of this session: look at whats made, draw over my feedback, then send it back. I did this a couple more times.

I think Claude models are better at inferring intent from design work where GPT is more rigid so it needs more specific feedback for tweaks.

Then I tried a design-review skill made by Emil Kowalski, a designer whose work I like. I generally don’t like agents calling a skill unless I specifically ask for it so I have my agents disable that on all my skills.

Just say something like:

Then a few more back and forth tweaks. But it got to a pretty good place I think.

Putting the content in files

I wanted every thread on the site to get its content from it’s own Markdown file so I could easily change content in those files instead of trawling through a bunch of code.

And I wanted everything saved in a folder ~/repos/bentossell - but I forgot I already had a project for my current personal website, but I hadn’t said that… The agent found the existing project and started creating a new version around it.

But I wanted to just replace the old with the new.

Then I went mobile (kids!) so wanted the site deployed to a url that I could visit. I use here.now for most of my quick builds - anything that isn’t a big app with data moving in/out, a database and things like that. I’m an investor in the company…but this is not sponsored. I invest in tools I want to use.

But you just give your agent the instructions, it installs a skill then publishes any sites for you. And you can of course get a paid account for persistent urls and custom domains. More on that later.

Like my manic agent session walkthroughs?

Share

Scope-creep

I ran another design review, this time focused on animation, and asked the agent to address the findings.

Then made a to-do file and grouped ideas by complexity. Some were minor interface fixes. Others changed the project: add a real AI model, let visitors send messages, show real thinking and tool calls, and create artifacts or canvas previews inside the site.

I had Codex use a different agent, Droid, to run some of the work with Grok because I was not getting what I wanted from the GPT models on the interface. When it finished, I asked it to save (commit) and redeploy everything.

By about 10pm, I had a long list of things the site could become. It could let people install a skill. It could contain real agent chats. I started creating an animated pet, my dog Nero, that could be dragged around and animate. Scope creep.

I slept on it and hated the whole thing.

It was too complicated for a personal site. If I made it look like a real agent, I would want to make it a real agent. There would always be another button, interaction or easter egg to add.

The concept is something I love but I’ll save it for another build in the future. So I scrapped it.

I often don’t know what I want until I can see and use a version of it. This is why my design instructions usually ask for three versions. I need to feel the differences. Sometimes that shows me the right direction; sometimes it shows me that the whole direction is wrong. Both equally helpful, usually 😅.

Starting again with a canvas

Before the agent-interface idea, I had thought about making the site feel more like a canvas.

You’ll have noticed I’m using tldraw a lot for explaining my sessions so thought I could make my website more ‘canvas-y’.

I made a rough mock-up with handwritten notes and small previews. Hovering over a piece of text could reveal an arrow or annotation.

I’d seen a personal site (shoutout Daniel White) that used a file-tree structure as a timeline. It showed current and previous work as nested items. That’s cool, let’s ‘borrow’ it.

It made three versions of the new direction. Let’s just say I wasn’t that impressed.

Then I got back into screenshots and annotations tweaking small design implementations and how it felt (should things auto-expand, be collapsed, smoothness of the open/close etc).

I moved all the page content into Markdown again, then removed the old agent-interface files and the unused animation frames for the dog (sorry Nero!).

Putting it on my domain

I deployed (published) the simpler version with here.now. This time I wanted it on bentossell.com rather than a temporary address. My domain is managed through Vercel, so I told the agent to use Chrome, add the domain in here.now and update the DNS records in Vercel.

The agent handled the setup and the site was live. It took about 20 minutes for the domain change to settle, which is normal when a website starts pointing somewhere new.

I checked whether here.now needed the project to be on GitHub. It didn’t. The agent published the files directly from the local project folder to here.now. It creates a site, uploads the local files and then makes that version live.

I still used Git for version history. A commit = a saved version. But if you ‘push’ those commits, they get sent to your GitHub repo. Useful for keeping a copy of the code and seeing how it changed, but not needed for this deployment.

Giving the simple site some character

The new site was deliberately more basic, so small details made a bigger difference. I drew a little mascot in the canvas and had seen another personal site with a character bouncing across the top of the page. I wanted mine to waddle around in the same way.

This is from Noah’s site.

The first animation looked rough. Its eyes needed to sit behind the rest of the face. The arms disappeared. The image flipped at the edge of the screen, which looked strange, so I asked it to walk sideways like a crab instead. I adjusted the legs, size and shadow, then asked the agent for ways to improve the basic animation.

I added keyboard controls. The up arrow makes the mascot jump; left and right make it move faster. Hovering reveals a small tooltip, and clicking it does something useful rather than treating the character as decoration. I also used the mascot as the site favicon: the small image shown beside a page title in a browser tab.

I still liked the annotations from the first canvas idea. Rough Notation is a small library that draws animated underlines, boxes, circles, highlights and brackets around parts of a web page. I had seen another site use brackets with handwritten explanations and wanted to try the same thing.

I drew over another screenshot to show where a bracket should sit and what it should explain. One marks the Markdown formatting. Another points out that a section can expand. On a smaller screen, the text has to move below the content instead of being squeezed beside it

The first responsive version still had problems. Expanding a section was faster than its bracket animation. Some brackets stretched far beyond the content. I annotated those details and asked for the bracket to stop at the content width. I narrowed the page, cleaned up the animation and deployed again.

Later that evening I added light and dark themes. I normally use one icon to switch between them, but then I questioned whether the switch needed to exist at all. I removed it and let the site follow the visitor’s system setting.

Where it ended up

The current site is much simpler than the first one. Content lives in files I can edit. The tree can expand, the annotations move with it and the mascot wanders across the top.

There’s still things I want to add. The difference is that each addition now has room to matter. In the agent-interface version there were so many controls that I could have spent forever adding Easter eggs and interactions without improving what the site was for.

A lot of the session was small back-and-forth work on the interface.

That is how the site was rebuilt: one complete direction, a night away from it, then a second direction worked out through screenshots, annotations and many small changes.

See for yourself - bentossell.com

I know many of you are shouting ‘the agent design was way better’ - yeah, I know. But as with every previous version of my personal site, I’ll pretend to update it regularly and then leave it for dust. The agent interface I want to use in another project and spend an inordinate amount of time and tokens making it great.

If you’re that annoyed, I encourage you to make yours an agent interface and show me 😊

The point of me sharing these sessions is because I’m still learning but I know enough to build pretty much anything I want to (it may take some time and swearing). And I want to show that I’m not doing anything special to get these agents to do stuff for me. It’s just figuring out what things are.

Get good at the thing while doing the thing.

The last few posts have been pretty basic stuff but I’ve got much bigger builds in the works and I plan to keep sharing what I build and how the digital sausages are made.

Behind the scenes

Enjoy your weekend!

Feel free to comment with any questions.

Subscribe now

• Find me on X

• Read about me and Ben’s Bites