I'm going to give Droid a shot. I had so far been avoiding it because it felt "too technical" but given you're just like me (not an engineer, but understands deeply product development concepts, technical concepts, etc.) this may just be the push I needed.
Have you tried Antigravity? I love the inbox feature, but I assume you may dismiss it as too basic if you're using a tool like Droid so heavily.
i used it when it launched but that was it. I may do some sort of workshop next year + posts on how i actually / practically do this. it looks messy but theres a lot of nuggets in there imo
Would love that. I want to build out my flow and setup so that I can be really intentional about when I commit from an idea to a prototype. Feels like I'm just mashing a lot of buttons at the moment (which kinda works!) but with young kids and a full time job I want it to be tighter.
Hi Ben, interesting post and thanks for the insights.
I’ve been seeing a structural failure in how AI and operations intersect in real-world workflows: specifically in how medspas lose inbound client intent when human availability drops and normal systems aren’t designed to capture, route, and resolve demand.
It’s not a tool problem but an infrastructure problem, and I think it ties into broader discussions about how AI should be developed as infrastructure rather than one-off applications.
If this is useful for the community, I’d be happy to share more context and patterns.
Hi Ben, just wanted to let you know that your subscription form seems not to be working. We input the email and press Subscribe, nothing happens. Have a great New Year and thanks for all the great content!
Great wrap up, Ben! I have enjoyed reading your email newsletters over the last months. Quick question: What feedback signals improve agents fastest without introducing instability?
Thanks for considering my question, Ben! I will try and break it down further.
How does an AI agent learn what “good” looks like over time? - given the autonomy they enjoy. Also are there any feedback signals that allow AI agents to improve & go in the right direction?
i think this will happen soon, as rn you have to setup self improving systems like if you run into issues you can then update agents.md with instructions for the next time
Your newsletter is the best one in Tech. Only thing missing is a dark mode option. The white bright background makes it very hard to read in the email. Please, if possible design a dark mode email.
Great job with this newsletter this year. Really appreciate the content and effort you put into it!
ty rick!
🙏
Expanding the builder base is fantastic. We call it the Age of Productivity: https://amicusai.substack.com/p/dont-fear-the-opex
I'm going to give Droid a shot. I had so far been avoiding it because it felt "too technical" but given you're just like me (not an engineer, but understands deeply product development concepts, technical concepts, etc.) this may just be the push I needed.
Have you tried Antigravity? I love the inbox feature, but I assume you may dismiss it as too basic if you're using a tool like Droid so heavily.
i used it when it launched but that was it. I may do some sort of workshop next year + posts on how i actually / practically do this. it looks messy but theres a lot of nuggets in there imo
Would love that. I want to build out my flow and setup so that I can be really intentional about when I commit from an idea to a prototype. Feels like I'm just mashing a lot of buttons at the moment (which kinda works!) but with young kids and a full time job I want it to be tighter.
Hi Ben, interesting post and thanks for the insights.
I’ve been seeing a structural failure in how AI and operations intersect in real-world workflows: specifically in how medspas lose inbound client intent when human availability drops and normal systems aren’t designed to capture, route, and resolve demand.
It’s not a tool problem but an infrastructure problem, and I think it ties into broader discussions about how AI should be developed as infrastructure rather than one-off applications.
If this is useful for the community, I’d be happy to share more context and patterns.
Dan Ouellet | Founder, Neurovia | https://accessneurovia.com | Vancouver, Canada
Hi Ben, just wanted to let you know that your subscription form seems not to be working. We input the email and press Subscribe, nothing happens. Have a great New Year and thanks for all the great content!
Hey, it shows that you're subscribed to the newsletter. Were you trying from another email?
Great wrap up, Ben! I have enjoyed reading your email newsletters over the last months. Quick question: What feedback signals improve agents fastest without introducing instability?
ty! wdym by feedback signals to improve agents?
Thanks for considering my question, Ben! I will try and break it down further.
How does an AI agent learn what “good” looks like over time? - given the autonomy they enjoy. Also are there any feedback signals that allow AI agents to improve & go in the right direction?
i think this will happen soon, as rn you have to setup self improving systems like if you run into issues you can then update agents.md with instructions for the next time
Your newsletter is the best one in Tech. Only thing missing is a dark mode option. The white bright background makes it very hard to read in the email. Please, if possible design a dark mode email.
Thank you for the great work
thanks - think thats a substack/your email provider issue - not something i can change