Hi all - Jeremy Howard from Answer.AI here. Really excited to share with you all what we've learned over the past year about how to work with AI in a way that's entirely human-centered.
Whilst most folks seem focused on how to remove humans from the loop entirely and make AIs do all the work, we've concentrated 100% on how to make humans part of the loop in a way that makes us more and more capable and engaged.
I've enjoyed building and using our tool, "solveit", for the last year, and do basically all my coding, writing, reading, research, etc in it nowadays. I use small fast iterative steps and work to maximize my learning at each step.
Hey everyone, Eric Ries here. solveit is the AI environment I personally have been using every day for months, not just for code but for writing and research, too.
it’s solved all the problems and frustrations I’ve had with both vibecoding and the limitations of the chatbot interface for doing deep work that requires concentration + the ability to understand the artifacts you are producing
and, as a special bonus, people in this course will get a sneak preview of the new book I’m working on. we’re going to use it both to teach some of the concepts from it (on how to create mission-driven long-term companies) and how to use solveit for longform writing projects
happy to answer any questions here, for folks that want to learn more,
I read most of this before understanding that I wasn't reading about some new agent or IDE or something, I was reading a sales pitch for a coding course for would-be vibe coders, with AI training wheels in the form of ... a dialog box to talk to an LLM.
I should have noticed the camp counselor / cultish / tedx vibes, throwin around REPL and feedback loops. I feel that it's somewhat misleading to present this as some amazing self-building software or server platform here, and bury the lede that what's being sold is an experimental tutoring method. It's almost like those "I built an AI agent that builds AI agents" posts, only instead of selling the sixty lines of python, it's selling a set of lectures that goes with them.
I'm sure I'm not the only one confused by this, but can you give details on why you decided that a course was necessary to learn this new way of working with AI?
Maybe it's more of a alpha thing, but with millions using chatbots every day, was it not possible to develop a UI?
It's not just about adoption, who has the time to spend 5 weeks learning a new tool? Particularly when you're competing with the existing tools?
Without the intent of hijacking whatever it is you are trying to achieve: I've found the best antidote to the AI-fatigue is to rely less on it. There is no way I am going to spend my day, or my employees day for that matter in reviewing thousands of LoC of bad AI-slop. In my teams we've dialed it back to just consulting mode and asking suggestions,e.g. replacement for good old search, because once you ask the GenAI to write a few thousands of LoC for you, you're also abandoning a lot architecture decisions (which you then have to figure out ad-hoc again, when you do the review of the slop, or at the latest when you notice the "smart" tool has yet again put a secret in plaintext or something similar). So yeah, if you have the so-called AI-fatigue, just use less of the so-called AI.
One thing I loved about the first solveit course was how create the community is. It goes back to fast.ai too, but everyone is super kind, smart, and has diverse backgrounds.
That testimonial page looks super fake. Why does the same Mathew Miller have 7 testimonials? Similarly, Pol Alvarez Vecino, Pierre Porcher and Duane Milne all have 6 testimonials. And almost all other users also have multiple testimonials. This can't be organic.
I thought it might be helpful to post a link to one of my favorite writeups from the beta cohort for solveit (last year). It's written by Chris Thomas:
Being among the first 1000 people to experience SolveIt has felt like witnessing the early days of a significant shift in how we work with AI. As someone who is a seasoned programmer, I have seen many programming paradigms and the advance of AI coding tools. What makes SolveIt different is not just another tool or framework - it is a fundamental rethinking of the human-AI relationship.
As I look at my experience with SolveIt, I think this is a better more sustainable approach to AI-assisted development. The current trend of ever more powerful models generating ever larger blocks of code feels unsustainable. SolveIt offers a different path. By maintaining human agency, working in comprehensible increments and building genuine understanding at each step, it creates a positive feedback loop where both human and AI capabilities grow stronger over time. This represents a partnership model that builds competence over time rather than creating dependence.
The implications extend far beyond programming. Whether I am implementing computer vision algorithms, exploring culinary science, or writing technical articles - the same principles apply. Small steps, continuous understanding, iterative refinement and always keeping the human as the agent in the process.
This is such a steep price tag. I loved what Jeremy Howard put up on fast.ai and respect the heck out of his team, but I've seen too many people scammed by online courses that sell a dream. This one seems to be selling a dream as well.
I'll be purchasing the course to try it out but I think my concern is not a one-off thing.
Why does this article not mention what solveit is at all? It talks about what they did, then that they made this tool, then that it's great, but what is it? Watch this video!
No, give me a sentence or two about what it does. I'm not watching a video about a tool while reading a blog post about it because you couldn't be bothered to write a line or two about it.
I'd urge folks here to atleast go through https://www.youtube.com/watch?v=DgPr3HVp0eg before jumping to conclusions about what SolveIt or this course is about. Its the polar opposite of vibe coding IMO.
I couldn't love more all the intentions behind this, but I have no idea what it is. Why would someone who already knows and loves Polya and iterative programming and human-centered technology... use this? What is the value add?
As a self taught hobbyist I progressed pretty far in advent of code 2023 until I gave up and less so in 2024 but my approach seems to be close to the one described (if you dig a little deeper into the signup page) or so I imagine. I was disciplined about not asking for help with the problem itself and went to chatgpt for help with components or syntax I needed to build a function I already had in mind (which was closer to the state of the art - especially in 2023 anyway). I think the advent of code problems are really interesting and have enjoyed solving them and watching others solve the ones I couldn’t. They are a fun way to frame the course. However the real value to me is learning how to approach more ambiguous problem spaces. I am definitely interested.
I had access to GitHub Copilot as a student in early 2022 while learning Haskell and immediately realised that it would hinder my learning if I didn't turn it off and implicitly follow this understand, plan, execute, reflect loop.
AI products like Cursor have the notion of an 'autonomy slider' [1] that can fortunately be turned all the way down (disable Cursor Tab) but relying on this discipline seems fickle when with the right agentic loops [2] and context engineering, thousands of lines of code can be churned out with minimal supervision.
I've considered always working on two projects over a long timespan, one with no AI assistance, possibly in a separate IDE like Zed, and one in Vibe Kanban (my current daily driver) but this feels like an inefficient proxy to accelerating this four step learning loop with a tool like solveit.
Since the solveit product isn't released and seemingly isn't competing with solutions, is there an opportunity to convey how AI product developers should be thinking about amplifying their users and keeping them in the learning loop?
So far, I've seen Claude Code's Learning output style [3], and also ChatGPT's study mode but in these cases, the only product change is a prompt and solveit is more than that.
I apologise folks that we did a bad job of explaining exactly what we're launching! My bad. :( I've added this to the top of the article now -- I hope this does a better job of explaining things:
----
*tldr from Jeremy:* You can now sign up for Solveit, which a course in how to solve problems (including coding, writing, sysadmin, and research) using fast short iterations, and also provides a platform that makes this approach easier and more effective. The course shows how to use AI in small doses to help learn as you build, but doesn't rely on AI at all -- you can totally avoid AI if you prefer. The approach we teach is based on decades of research and practice from Eric Ries and I, the founders of Answer.AI. It's basically the opposite of "vibe coding"; it's all about small steps, deep understanding, and deep reflection. We wrote the platform because we didn't find anything else sufficient for doing work the "solveit way", so we made something for ourselves, and then decided to make it available more widely. You can follow the approach without using our platform, although it won't be as smooth an experience.
The platform combines elements of all these: ChatGPT; Jupyter Notebook + nbdev; Bits of vscode/cursor (we embed the same Monaco editor and add similar optional AI and non-AI autocompletion); a VPS (you get your own persistent full VPS running Linux with a URL you can share for public running applications); Claude Code (all the same tools are available); a persistent terminal. Then there's some bits added that don't exist elsewhere AFAIK: something like MCP, but way simpler, where any Python function can be instantly used as an AI tool; the ability to refer directly to any live Python variable in AI context (but optional, so it doesn't eat up your context window); full metaprogramming of the environment (you can through code or AI tools modify the environment itself or the dialog); context editing (you can -- and should -- directly edit AI responses instead of tell the AI it's wrong); collaborative notebook coding (multiple people can edit the dialog, run code, etc, and all see live updates).
I participated in the first cohort and I'll be doing it again because I enjoyed it so much the first time around. The course focuses on teaching a robust problem solving approach, rather than explicitly teaching people how to program with AI. It's not a dream or a scam! If you digest the course concepts, the takeaways can be put into practice even if you aren't programming directly in SolveIt. But the SolveIt application certainly greases the wheels by making this problem-solving approach easier and fun! My growth as a programmer has been supercharged by what I learned and applied in the past year since taking the first course.
A great course and helps with learning proactively and not reactively. Important in this age of ai. I was part of cohort 1 and have enrolled for the next one.
34 comments
[ 5.7 ms ] story [ 67.1 ms ] threadWhilst most folks seem focused on how to remove humans from the loop entirely and make AIs do all the work, we've concentrated 100% on how to make humans part of the loop in a way that makes us more and more capable and engaged.
I've enjoyed building and using our tool, "solveit", for the last year, and do basically all my coding, writing, reading, research, etc in it nowadays. I use small fast iterative steps and work to maximize my learning at each step.
it’s solved all the problems and frustrations I’ve had with both vibecoding and the limitations of the chatbot interface for doing deep work that requires concentration + the ability to understand the artifacts you are producing
and, as a special bonus, people in this course will get a sneak preview of the new book I’m working on. we’re going to use it both to teach some of the concepts from it (on how to create mission-driven long-term companies) and how to use solveit for longform writing projects
happy to answer any questions here, for folks that want to learn more,
Eric
Side note: supposedly this is the first cohort of this course, so how do you already have testimonials?
I should have noticed the camp counselor / cultish / tedx vibes, throwin around REPL and feedback loops. I feel that it's somewhat misleading to present this as some amazing self-building software or server platform here, and bury the lede that what's being sold is an experimental tutoring method. It's almost like those "I built an AI agent that builds AI agents" posts, only instead of selling the sixty lines of python, it's selling a set of lectures that goes with them.
Maybe it's more of a alpha thing, but with millions using chatbots every day, was it not possible to develop a UI?
It's not just about adoption, who has the time to spend 5 weeks learning a new tool? Particularly when you're competing with the existing tools?
We've captured a slice of that on our main site. Testimonials: https://solve.it.com/testimonials Some blog posts: https://solve.it.com/#showcases on the main page
And on of the students even made a project dashboard page showcasing all the things everyone has built! https://solveit-project-showcase.pla.sh/
He even blogged about it : ) https://himalayanhacker.substack.com/p/how-i-built-solve-it-...
https://christhomas.co.uk/blog/2025/09/24/the-human-is-the-a...
a few quotes/excerpts:
I'll be purchasing the course to try it out but I think my concern is not a one-off thing.
No, give me a sentence or two about what it does. I'm not watching a video about a tool while reading a blog post about it because you couldn't be bothered to write a line or two about it.
AI products like Cursor have the notion of an 'autonomy slider' [1] that can fortunately be turned all the way down (disable Cursor Tab) but relying on this discipline seems fickle when with the right agentic loops [2] and context engineering, thousands of lines of code can be churned out with minimal supervision.
I've considered always working on two projects over a long timespan, one with no AI assistance, possibly in a separate IDE like Zed, and one in Vibe Kanban (my current daily driver) but this feels like an inefficient proxy to accelerating this four step learning loop with a tool like solveit.
Since the solveit product isn't released and seemingly isn't competing with solutions, is there an opportunity to convey how AI product developers should be thinking about amplifying their users and keeping them in the learning loop?
So far, I've seen Claude Code's Learning output style [3], and also ChatGPT's study mode but in these cases, the only product change is a prompt and solveit is more than that.
[1] https://www.latent.space/i/166191505/part-a-autonomy-sliders [2] https://simonwillison.net/2025/Sep/30/designing-agentic-loop... [3] https://docs.claude.com/en/docs/claude-code/output-styles#bu...
----
*tldr from Jeremy:* You can now sign up for Solveit, which a course in how to solve problems (including coding, writing, sysadmin, and research) using fast short iterations, and also provides a platform that makes this approach easier and more effective. The course shows how to use AI in small doses to help learn as you build, but doesn't rely on AI at all -- you can totally avoid AI if you prefer. The approach we teach is based on decades of research and practice from Eric Ries and I, the founders of Answer.AI. It's basically the opposite of "vibe coding"; it's all about small steps, deep understanding, and deep reflection. We wrote the platform because we didn't find anything else sufficient for doing work the "solveit way", so we made something for ourselves, and then decided to make it available more widely. You can follow the approach without using our platform, although it won't be as smooth an experience.
The platform combines elements of all these: ChatGPT; Jupyter Notebook + nbdev; Bits of vscode/cursor (we embed the same Monaco editor and add similar optional AI and non-AI autocompletion); a VPS (you get your own persistent full VPS running Linux with a URL you can share for public running applications); Claude Code (all the same tools are available); a persistent terminal. Then there's some bits added that don't exist elsewhere AFAIK: something like MCP, but way simpler, where any Python function can be instantly used as an AI tool; the ability to refer directly to any live Python variable in AI context (but optional, so it doesn't eat up your context window); full metaprogramming of the environment (you can through code or AI tools modify the environment itself or the dialog); context editing (you can -- and should -- directly edit AI responses instead of tell the AI it's wrong); collaborative notebook coding (multiple people can edit the dialog, run code, etc, and all see live updates).