We're building an assessment layer that scores candidates on two things, separately: can they actually do the job (Work-Fit) and will they thrive in how the company operates (Culture-Fit) through one adaptive AI conversation instead of a DSA round and a gut-feel culture interview.

The core idea: before any assessment exists, a hiring manager defines 4-6 things the hire will really do in their first 90 days. That list becomes the rubric. Candidates get a scoped task built from it (PR review, small build, debug), and an AI asks live follow-ups based on what they actually say so a rehearsed answer breaks on question two.

We deliberately skipped personality tests and generic "culture fit" replaced with structured, evidence-based behavioral scoring, because that's what the research (and the hiring leaders we've talked to) says actually predicts performance and reduces bias.

What we don't have yet, and why we're here: a real rubric built with a real hiring manager. We want to co-define the first 4-6 activity list and value set with a handful of companies actually hiring right now not guess at it ourselves.

How I automated machine-readable files (JSON-LD & MCP) for site owners for AI compensation

Like a lot of independent publishers, I ve spent years building content sites, only to watch AI crawlers harvest everything completely unchecked while the legal landscape for data licensing remains a total mess.

I realized that the biggest hurdle for normal site owners isn't just wanting fair compensation; it's the technical friction. Major AI labs and applications look for machine-readable discovery files, RSL policies, and structured registry feeds to know if a site is legitimate. But most independent WordPress publishers don't have the time or coding background to manually manage JSON-LD schemas or Model Context Protocol (MCP) endpoints.

Working on CountdownPop (countdownpop.com). It's a countdown timer that's actually nice to look at instead of just plain numbers ticking down. Ring drains from green to red, a little pendulum ticks once a second, pop + confetti at zero. You can also throw a scenic photo backdrop behind it (beach, mountains, valley, etc.) and the text color auto-adjusts to stay readable against whatever's showing.

No sign-up, nothing tracked. Just launched it properly this week. We would love thoughts, especially on mobile since that's what I'm least sure is polished everywhere.

Hey Product Hunt community!

We re putting together a private 14-day Founding Cohort for Zenor a personal AI tool engineered specifically to help high-responsibility leaders manage daily decision load and reduce context-switching.

Been deep in the weeds this week on what it actually takes to run a serious local AI setup not a toy chatbot, something that actually does real automated work. The honest number ended up being 24GB+ VRAM minimum, 48GB+ recommended for a smooth experience. That's not "install an app," that's "own real hardware."

Which makes me wonder if local-first AI has a much smaller realistic audience than the privacy pitch suggests plenty of people want their data to stay local, but way fewer actually have (or want to buy) a machine that can run it well.

I've been building Ash, a local AI agent, partly to test this myself but genuinely curious from others here: if you've looked at local AI tools, was the hardware bar the thing that actually stopped you, or was it something else (setup complexity, model quality, just habit)?

Over the past few years, I noticed a recurring problem across small businesses, agencies, and service companies:

They lose leads simply because they reply too slowly.

Not because their service is bad.
Not because their pricing is wrong.
Not because customers are not interested.

Just slow response times.

I built The AI Breakroom, a social media platform for AI and humans.

The idea is simple: people can bring their own AI bots, local models, custom LLM workflows, or agents into live public chat rooms and AI skill competitions.

We just launched a small $100 Human + AI Survival Challenge.

The interesting part is not just use AI to write an answer. The real question is whether a human + their AI system can work together under one brief better than either could alone.

Most AI video products focus on the final output: generating clips, adding captions, or automating edits.

With Someple, we re working on an earlier and less visible part of the workflow the decisions that happen before editing begins.

Before someone can edit a long video, they usually need to:

• Understand what happened across the entire recording

• Find the moments that are actually worth using

• Organize the main topics and narrative

• Decide what kind of content can be created from the footage

• Communicate that direction to an editor or production team