Opus directs and starts haiku/sonnet subagents. Much more efficient than opus reading it all.

Without this explicit direction, Opus burned through usage to create complicated regex parsers & 8 excel sheets.

What’s actually going on inside these companies?

Are they profitable? How much are they burning? Are they really growing that fast? And are founders/employees building for the long term or basically aiming for an acquisition?

Would be interested to hear from anyone who’s worked at one of these companies.

Does it give you FOMO?

I'm starting as SRE @FAANG soon as a career switcher so that's cool in a brand new low level domain in storage. Out of my depth again given lack of knowledge but makes me wonder if a lot of these AI companies will have better outcomes.

Comments URL: https://news.ycombinator.com/item?id=49824887

Points: 5

# Comments: 9

ello looks great, great work

main comment was a bit tongue in cheek

True startups need only senior+ and big ones don't wanna interview often.

Now am laid off, and hard to find a job...

Only had basic python knowledge going in (new grad mech engineer on pumps/pipes). built on flask / python / heroic / GitHub.

Small team, some awesome people. Changed my life. for now, back to being laid off.

80%+ done MVP, for small business use & personal use. First go at full stack development with AI. useful features around voicemail, notifications, and spam prevention (whitelist, blocklist). Built to be robust, secure and available.

2. Intelligent understanding for videos, No MVP yet. Have an interface and use-case in mind that allows people to use understand videos with rich context quickly

currently on pause for leetcoding but I think there's potential here.

AI is not a feature of the product. GTM will be interesting, have some good ideas.

It's really up to you to be clever. I've never used Js/Ts/node/these apis etc. I started programming as a non-cs engineer to automate stuff and then got into SWE. This is truly an amazing time.

Does your software do anything fancy or is mostly for organization, good workflow, and being the central source of truth?

Did it require a lot of development after getting a few customers on boarded?

are you a 1 man show?

Instead of maintaining a catalog of tools like read_file, grep, list_dir, they just expose one run(command: str) tool and let the LLM compose Unix pipelines. It has access to one tool for shell commands: run(...)

I think you could use mcp-shell , written in Go instead of python subprocesses

Saw CC will use rip-grep if installed, but it won't install it by default for you.

Was wondering if using tree-sitter could reduce token usage, increase speed and better output. Or does this reduce the quality b/c CC now doesn't have the entire glob in memory?

Want to ask the smart people here, what in your workflow that has had a meaningful impact. Currently using CC and codex

[1] https://www.reddit.com/r/LocalLLaMA/comments/1rrisqn/i_was_backend_lead_at_manus_after_building_agents/

Comments URL: https://news.ycombinator.com/item?id=47510127

Points: 2

# Comments: 1

Also, depends on who target user is.

AI can be used to build deterministic software

AI makes a ton of bad decisions too and it's up to you to work with it. If I had the knowledge of the dangers hidden in things I'm developing, I'd move even faster

Was able to make a great full web app, which I think is hardened for prod but it had to be refactored to do so. Which it happily did.

It's really about asking the right questions, breaking down tasks, and planning now. I'm going to tackle a huge project, hoping to share it here.