blancotech

Stumbled upon this and found it to be one of the most entertaining ways I've seen to explore a company/product’s history

> The hardest part of QA still seems to be maintaining tests as the system evolves

That’s interesting. It may explain why so many companies now push “self-healing” tests with LLMs for small UI shifts. The teams I spoke with faced different challenges, so the toughest part varied by where they stood in their QA cycle.

> Curious whether you’re aiming...

I started with a broad “AI test everything” approach, but I learned fast that the intent problem I mentioned is tough to beat. The prototype looked great in demos, yell fell short when I dogfooded them on my other projects. And when I met with teams, I didn’t see clear market pull. What comes next is still open.

Hey everyone, I’ve put together this blog post summarizing what I’ve learned over the past month while building a QA-focused startup. I’m still developing my understanding of the space, so I’m also looking for gaps I might be missing. All feedback is welcome!

Bundling and distribution. OpenAI has more paid subscribers than Anthropic. I started using Codex over Claude because Codex is included in my subscription.

It’s the classic Microsoft Teams vs Slack debate.

I’m curious how any project management to code agent workflow can be successful given how messy the process is in real life.

Especially discovering unknown unknowns that lead to changes in your original requirements. This often happens at each step of the process (e.g. when writing the PRD, when breaking down the tickets, when coding, when QAing, and when documenting for users).

That’s when the agent needs to stop and ask for feedback. I haven’t seen (any) agents do this well yet.

> An important mechanism here is that even if a crawler doesn’t have a billing relationship with Cloudflare, and thus couldn’t be charged for access, a publisher can still choose to ‘charge’ them. This is the functional equivalent of a network level block (an HTTP 403 Forbidden response where no content is returned) — but with the added benefit of telling the crawler there could be a relationship in the future.

IMO this is why this will not work. If you're too small a publisher, you don't want to lose potential click-through traffic. If you're a big publisher, you negotiate with the main bots that crawl a site (Perplexity, ChatGPT, Anthropic, Google, Grok).

The only way I can see something like this work is if a large "bot" providers set the standard and say they'll pay if this is set up (unlikely) or smaller apps that crawl see that this as cheaper than a proxy. But in the end, most of the traffic comes from a few large players.

Anyone else immediately think of delta airlines? I was excited to read an analysis of a seat-to-seat chat implementation

[x] did it just for fun

[x] instantly usable

[x] no sales pitch

These are the posts that keep us on HN

Nice idea but the results weren't comprehensive for the book I tested. I find that chatgpt with web search enabled does a better job with the simple prompt "Create a relationship diagram for the characters in <BOOK TITLE> using mermaidjs. Include the relationships between people."