Mohi Uddin

The biggest issue right now is honestly the debugging tax. We all know Cursor and Claude are goats for fixing bugs, but it’s actual pain when you hit the limit and realize you wasted all your tokens just trying to get the code to work. Even the paid users are getting screwed by rate limits and time-outs... but the free tier? If you’ve got a massive codebase, you’re basically cooked. #BuildingInPublic #AI #Debugging

but for what? images? dont say for coding!

Is anyone still using Gemini?

When you got tons of knowledge or information on your table...how would you manage it without losing a single thing?

There will definitely be changes in AI, but it's getting addictive...rn everyone is just shipping code without actually learning the core skills or trying to solve problems on their own... Once the entire dev community reaches a point where they literally can't code without assistance, big tech will absolutely jack up the costs for their frontier models. We're setting ourselves up for massive vendor lock-in. lets see

For a short while (last year or so) the vibe coding tools were useful for two reasons: 1. Better visual sense 2. Handling the headache of deployment and hosting (I mostly see them as hosting services) Opus and Sol have bought caught up on design sense and ChatGPT now has sites for hosting. If they just add domains and scalable plans then I’m not sure what advantage vibe coding tools have.

@waitin4agi_ What about inference cost? How long do they just suffer the cost and maintenance? What does the future work along with humans look like?

@poetengineer__ W work, seeing all these urges to learn the process and giving a shot....but you've mastered it...

what does ai actually add to open-ended creative work? my first attempt at giving it a process signature: from a real project's transcripts i pulled 146 ideation events; the cubes are the moments ai forked the ideas open, laid along a 3d chronological trace of the sessions.

@AndrewYNg @coursera Nowadays, we just give notes or a syllabus to Claude or, rarely to NotebookLLM, idk why no one is using it...but mainly we use Claude for learning n prep

Fifteen years ago, @Coursera and online courses changed education. It worked better than almost anyone expected, expanding access by opening up where you can learn. But how you learn remains largely the same as it has for centuries: it is still one-size-fits-all, taught the same way to each person who shows up. We now have an opportunity to change how learning happens. With advances in AI, we can now build a custom learning guide for each person. We will turn learning from one‑to‑many to one‑to‑one. I'm starting LearnVector to invent this next generation of learning. We are starting with a $100M investment from Coursera, and plan to collaborate closely with Coursera and Udemy. Good learning needs much more than just a chatbot. Research shows that chatbots without guardrails harm learning. They help complete tasks and enable students to do better on homework. But cognitive offloading to a chatbot results in them being less skilled. And, you cannot always trust what a chatbot tells you. In contrast, LearnVector will plan a path with you, adapt to how you learn, and patiently stay with you until you’ve mastered new skills. One thing has not changed in all this time. People want learning they can trust: material that is accurate, relevant, and worth the effort you put into it. Anything less wastes the most valuable thing a learner has: time. Coursera has a trusted library of materials from authoritative sources. LearnVector plans to work with Coursera to bring this trustworthy learning to everyone. I'm grateful to Greg Hart and the entire Coursera team for supporting LearnVector. I look forward to working with our talented team to change how we learn, and accelerate human development. learnvector.ai

@poetengineer__ thats really great!

as an artist my main tools are: coding and a software called touchdesigner (if you dont know the latter: it's a node-based gui thingy where you can do real-time visuals). im constantly baffled by the question why coding vs. td feels that different: td is more akin to sketching and coding is...typing. so i made an instrument that watches me work in td: every drag, wire, and delete, timestamped:

@Cristiano the goat!!

Pai feliz ❤️

did a new design of gif, lmk how is it, can i do better? and try it out : debugai.io/start?utm_sour…

shipped this on the MCP server too. took way longer than it should have. turns out "just paste your API key" is a bad idea when that key ends up in a config file that gets committed. fixed it properly instead of shipping the shortcut.

I was like, who was this guy, but after watching this video, salute to this man bro..the words, the feelings, I can't express it

@Built4human yup, thanks for it

@built4us man, exactly.. mindlessly dumping raw logs into chat is a terrible waste of tokens n limits, targeting the upstream origin programmatically is very smarter than just blinding doing back and forth guessing.... glad someone finally tackled this loop!!

setting up an MCP server used to mean: find the config file for your client, paste a JSON blob, go find your api key on some website, paste that in too, restart, hope. we deleted all of it. npx -y @debugai/mcp setup browser opens, you confirm a code, done. every client it recognizes gets configured automatically.

Yeah, that's the reason I built this, yk we build new features and hit a lot of errors, and we just used to copy n paste to other AI tools or just to our AI tool where we are working on and we just keep copy-pasting things saying " got this or fix this "isn't it? and burn a lot of tokens, and eventually hit the limit and in the coming times, due to inference cost, every single one will think twice before burning tokens

@built4us Nothing worse than chasing a runtime crash, pasting stack traces into Claude, and getting hit with 'message limit reached' right as you're halfway to the fix. If this actually traces the upstream cause right in the editor, will def giving a try...

@thsottiaux He won't stop giving resets.

Update. I have decided to take a break from x to recharge a bit. See you back tomorrow for more ChatGPT and Codex fun. It's going to be a fun week.

not claiming this replaces reading your own code. claiming it replaces the five minutes of grep before the reading starts. npx -y @debugai/mcp setup free tier, 10 debugs a day, no card: debugai.io/start?utm_sour…

the loop does not end at the fix. agent applies it, calls report_outcome, that result feeds team error memory. next time your project throws the same error signature, the fix that already worked leads.