Most founders treat categories like labels.
Product Hunt treats them like distribution.

Categories weren t added to classify products.
They were added because one global feed stopped working.
Too much noise. Too little intent.

Your category decides:

• who sees you

• how you re evaluated

• the quality of feedback you get

Most people saw AI startups. The real shift? AI as infrastructure.

~160 companies accepted. The signal was clear:

1. Agentic AI (~30%+)
   Not wrappers.
   Systems executing multi-step workflows autonomously.
   Replacing humans, not assisting them.

2. The vibe-coding edge (~25%)
   1 in 4 companies had ~95% AI-generated codebases.
   AI wasn t just a tool; it was the development process.
   Speed became the moat.

3. Vertical > Horizontal
   Generic productivity lost to domain automation.
   Tighter workflows. Clear ROI. Stronger defensibility.

4. Workflow automation (~15 20%)
   Hiring. Ops. Onboarding.
   Expensive, repetitive systems, now automated.

@Y Combinator didn t fund AI companies.

We kept hearing get your ICP right.
But what we learned is that who you reach out to first matters just as much as who eventually decides.

In most companies, there isn t one ICP. There s a sequence.

• Someone experiences the problem daily.

• Someone else prioritizes it.

• Another person signs off on it.

If you jump straight to the top, you often lack context.
If you stay too low, momentum dies.

I realized I was stuck in AI FOMO.
Bought multiple courses. Knew every tool by name.
Hadn t built a single working automation.

So I stopped and asked one question:
"What repetitive task can I hand off to AI today?"

Not after another course. Not after learning more. Today.

That shift mattered.

A lot of people read YC RFS Spring 2026 as a trend list.
It s not. It s a signal of where work inside companies is quietly breaking.

Here s how this shows up in real teams:

Product teams
YC references @Cursor , but the opportunity isn t coding faster.
It s helping PMs synthesize interviews, metrics, and feedback to decide what to build next.

Finance and hedge funds
Firms like Renaissance, Bridgewater, and D.E. Shaw won by systematising decisions.
AI-native hedge funds push this further with continuous, machine-driven strategies.

I looked into a few of their launches and what stood out wasn t a secret hack.
It was how little they tried to launch.

Their tagline isn t hype.
It s literal. Write, edit, and chat about your code with AI.

No buzzwords. No promises. Just what it does.

The pattern is simple:

This debate often gets framed as Should researchers use AI for literature reviews?

I think the real question is different.

Is it ethical to spend hundreds of researcher hours on mechanical work when that time could be spent advancing actual knowledge?

Think about a researcher spending an entire weekend searching papers, skimming irrelevant abstracts, copying citations, and fixing references. That s not insight or discovery. That s overhead.

As more teams build AI agents, search, and personalized feeds, one problem keeps surfacing.
Not generation.
Not model quality.

It s retrieval and ranking. Deciding what information should show up and in what order.

Most teams solve this by stitching together systems. Vector search for meaning. Keyword search for precision. Custom logic for business rules. Over time, relevance logic spreads everywhere and becomes hard to change.

@Shaped approaches this differently.

Yesterday, I had a chat with a founder who s launching on Product Hunt next week.

He said: We ve got a community of 500+ people. Getting Product of the Day should be easy.

So I asked one question:

How many of them have an active Product Hunt accounts that are at least 4 6 weeks old?