spgorbatiuk

There's clearly room for improvement there, you're right, but I'd also say these kinds of tasks are severely underconstrained, and if you have a certain image of how it should look in mind, you should give it a lot of references, especially if UI is complex. Screenshots, figma, sketches, references to existing design, etc. And even then there's probably going to be a few iterations.

I've had a pretty good experience tampering with existing UI without taking time of the frontend team, but depending on the scale and your abilities, some things may indeed be faster manually still

In short, so to not repeat the others' advice -- yeah, go find communities where those people you think will find this useful hang out. Contributing will not only spread the word, but also help you refine some of your assumptions for sure.

This can be time-consuming, so there's sometimes a tempting direction to try and distance from this process by using marketing automation tools that would post for you, but I would advice against that precisely because there is a lot of value to just be present where your potential users are. Using AI carefully to help you scout the most interesting topics and discussions, though, may be a remedy, and I myself sometimes do that. Good luck!

The value was never really in the building, it was just hidden by how expensive building used to be. When making the thing was the hard part, you could mistake "I built it" for "I have a business." Now that the build is cheap, that illusion is gone, which is uncomfortable but clarifying, and sighs building something OTHER people want is still arguably hard...

Frankly, coding with Claude Code and having Copilot read through the PR is complementary and helps to catch some things that slipped through

It would help if you could provide a bit more details on what you're building and in which domain.

But from what you provided, I would make sure you really get to the bottom of why this pilot happened with your customer: did they have a problem specific enough? What changed right before they came to you (a new hire, a new regulation, a system that broke)? Who felt the pain, and who signed? Write those down as attributes, and your ICP becomes "companies that match the trigger," which is often not the same as "companies in that vertical."

Yeah, my assumed neurodivergence has never been more apparent to me than in these moments. Sometimes it feels like genuinely more stuff is getting done in parallel, but I haven't figured out the key to this kind of performance. So, at times it is just the fatigue from context switching without the results that would justify it.

I don't know if it's just my perception, but I think the "channels" feature is massively underestimated. I get that it's basically their response to openclaw, and people kind of had enough of that stuff, but having my agent in my favourite messenger is a game-changer for me.

It's different for me from just app chat, because I have some specific custom workflows and tools I gave to the agent, and, yes, all of the nice stuff mentioned in other comments

We're a startup, and every role uses AI for many tasks, at times quite efficiently. There's no organised strategy rn, just the generous limits incentivising the use.

The only organised effort, I'd say, is on the engineering side with eventual creation of shared custom skills for coding agent that go into the repo.

Hardware and provider juggling is a way to go, although I think it is also worth mentioning that the cost is not only the price-per-token, but first of all, the amount of tokens used.

Depending on what one builds, comprehensive documentation and applicable skills and memory tools often allow for a substantial reduction of tokens previously used by the agent to comprehend and remember what is being built

In an unlikely scenario you haven't stumbled upon the superpowers (https://claude.com/plugins/superpowers), I really recommend the /brainstorming skill there. The plugin itself leans into engineering domain, but this specific skill makes claude think better and bring up and ask thoughtful stuff before it gets to a solution part

Sad, but fair!

Not sure if I got the question right, but there are benchmarks like SWE pro and stuff. There's whole another debate whether you can trust it or not, and whether the labs are training on those benchmarks, but that's one way to measure that.

Other than benchmarks, I'd say that's your own test suite

Now, that development and prototyping is way cheaper than before, how would you say an approach to ideas validation change? Which types of hypotheses are better and faster tested using via vibe-coding, and which should still be verified like we used to?

Both CC (mostly pet projects and automation), and Cursor (mostly at work, because I still read the code, interact with python notebooks, etc.)