howon92

This is a great idea. Can you share more about what "24 specialized agents" mean in this context? I assume each agent is not simply an LLM model with a specific prompt (e.g. "You're the world's best biologist. Review this biology research paper.") but is a lot more sophisticated. I am trying to learn how sophisticated it is

> To my surprise, Zoo’s API didn’t perform particularly well in comparison to LLMs generating STLs by creating OpenSCAD

This is interesting. As foundational models get better and better, does having proprietary data lose its defensibility more?

Training an audio model this good from 0 prior experience is really amazing. I would love to read a blog post about how you guys approached ramping up knowledge and getting practical quickly. Any plans?

Congrats! What made you not want to take any investments and what was the most counter intuitive thing that made you succeed?

Thank you! Interesting, because I haven't perceived "terminal performance" as a problem before. I will try it out

I'm a fan of your work! I'm curious about how you decided to work on building a terminal for your next project among your other ideas. If you have time later, could you share your main motivation with us or link to an existing post if you already mentioned it elsewhere?

To people who tried using it, what are the reasons to use it over iTerm2?

I love the landing page. Great job!

What problems did you have with Twilio?

Where does the data come from?

Wow this is really cool

Thanks for sharing your experience!

That IS the best answer.

My exact thoughts

I used to copy and paste code from ChatGPT to my IDE to render React components. This Chrome Extension helped me render the React components directly in ChatGPT and save me a step.

My feedback: keep the same style :)

I love it. Thank you for making it

There is nothing wrong with pivoting

> In general, I'm leaning towards a strategy where I freeze the product for a while and put 99% of the time + money + energy into marketing for purposes of finding PMF.

Have you looked into cohort retention of your users? If your users continue using your app, then I'd say keep going. Otherwise, there's no point in spending effort in marketing IMO.

> The best I've gotten is a link to an online leetcode-style test and I suck at those too

Get better at it by practicing more.

Well that’s my question. Adobe did exactly that to eliminate many companies and projects over night with Firefly

Thanks for the interview! All I think about these days is how I would build a moat in the AI space and was excited to learn what Youssef thought. This is a summary of that topic from the interview. Do you guys think this will actually help them build a moat around their product?

---

- Youssef argued their moat is focusing on the specific use case of podcasting and building a product tailored to that, rather than just being a wrapper around underlying TTS APIs. - While the APIs exist and can serve many use cases, they aren't specifically built to serve podcasting optimally like Wondercraft is.

- By being opinionated in their approach and selecting the best TTS engine for podcasting, Wondercraft aims to provide a better experience than what could be achieved through generic engines alone.

- They are building out templates and defaults tailored to different podcast styles (e.g. interview format) to simplify the process.

- Network effects from high podcast download volumes on platforms can also contribute to the moat by establishing Wondercraft as the go-to tool in the space.

- First-mover advantage in a new niche like AI-generated podcasting gives them time to refine their product before competitors emerge focusing on the same use case.

- Customization options and integration with other tools/workflows could deepen customers' investment in Wondercraft's platform over time.

Congrats on the launch!!! I suggest you highlight "why lantern is better than pgvector" at the top of your page. The first thing that came to my mind after reading this was "why should I use this instead of pgvector?"

No subscriptions, just $20: what a nice offering.

Many comments are criticizing the usage of LLM for this use case but I do believe this will become more common in the future. For example, OpenAI's retrieval plugin leverages LLM to do PII detection [1] instead of using the traditional libraries [2].

[1] https://github.com/openai/chatgpt-retrieval-plugin/blob/main... [2] https://github.com/topics/pii-detection

Congrats on your achievement! I'm curious about your end goal. Do you aim to beat GitHub Copilot's performance and convince devs to use Refact for code completion instead of GitHub Copilot? I want to understand the motivation behind these different code-completion models that are not solely for academic research.

This is a problem worth solving. Congrats on launch!

Move to a different environment where you feel like everyone else is more competent than you

I'm curious what the key advantages of AI Workbooks are compared to doing the work on Jupyter Notebooks!

This is a good chart. But I’d argue using Vercel is entirely justified for many indie devs or startups. I bet most won’t get to the point where the extra cost of using Vercel really hurts. If someone is lucky to get there, they can just switch to using Upstash / Neon if they feel the benefits of Vercel’s extra layer don’t outweigh the added costs.