Co-founder @meetaster - Conversational identity for builders. Your AI agent turns conversations into opportunities. Recovering recruiter turned builder.
I've been inside World Labs this year helping them hire across research and engineering.
Most of that time went on a small number of roles. The people who have actually trained frontier world models sit in maybe a dozen teams globally, so you spend weeks trying to reach a handful of them.
This week AMD agreed to buy the whole company for $8.2 billion. Ten weeks after World Labs bought SceniX to push deeper into embodied AI.
When the pool is that small, you can't recruit your way in. You buy the team. That's a good part of what the $8.2 billion investment is paying for.
So the scarcity isn't "AI engineers," it's people who have personally built and trained frontier models rather than worked near one.
Those two look identical on a resume.
If you're hiring into this pool, what's actually working?
https://t.co/TTKnNymFhE
Most candidates assume recruiters are reviewing their full background.
They aren't, and it’s because of what the software actually looks like. If you’ve never seen the recruiter side of LinkedIn, it’s eye-opening.
Your entire career, 10 years of craft, decisions, and hard-won wins, gets compressed into a tiny preview box.
A thumbnail, a logo, a job title, and two lines of text. That is all the interface gives a recruiter sitting in front of 500 applicants.
Crazy right? When software compresses a human being into two inches of text, pattern matching is the only way to survive.
If you could show a recruiter one piece of work that never fits in that preview box, what would it be?
I've sat in on a lot of calls where teams review profiles. The same thing happens in almost all of them.
Every few minutes, someone fills a gap with a guess.
No promotion in three years? Not ambitious. Manager, not head of? Wasn't really running it. One year in two roles? Jumpy. Thin profile? Doesn't take their career seriously.
Every one of those is an invention. It's what you do when you have to decide and the document doesn't contain what the decision needs.
And the guess always goes against the candidate.
Nobody sees a two-year gap and assumes it was someone's best work.
Very few jobs involve disappointing as many people as recruiting does. Most of the work is rejecting the 99.99% to get to the 0.01%.
Roles receive thousands of applications. Candidates fail technicals in the hundreds. Hiring managers routinely expect to meet 30+ people for a single opening.
Everyone loses in that arrangement.
Candidates go through a recruiter screen, a call with the hiring manager, a panel, and a weekend take-home project. Six weeks of thinking about your company.
Then silence.
When people get ghosted, they don't apply again. And they tell their network in group chats, DMs, and over coffee.
If you're hiring: what would change about your process if you had to write every rejection yourself?
15 years of interviewing people for a living. Last week the thing interviewing me was the product I'm building.
My agent, Aster, asks you questions so it can represent you. Mine asked better ones than I'd have written, and I answered honestly because nobody was assessing me.
Nobody believes that where you worked is the best way to judge what you can do. We use it anyway, for a boring reason: it's the only thing that survives being passed between five people.
Assessing someone properly is expensive. A logo costs nothing to forward.
Running a profitable business for 15 years without outside capital teaches you one thing: running a company without a deeper purpose is just a slow form of burnout.
No board. No VC safety net. Just carrying every market shock and operational hit alone.
In sport and entertainment, talent has an agent whose whole job is the best deal for their client.
In recruiting, it's a gray area. Today, no systems truly advocate for talent.
Is a human even the best advocate in hiring, or is AI better built to represent your actual work?
Web designer > UX designer > product designer > design engineer.
Prompt engineer got 18 months. AI engineer is doing four jobs and will split. Then a spec asks for 5 years in a title that's meant three things in 5 years.
Do the work early and you get filtered out.
Political science degree, portfolio years out of date. Any screen would have cut him, including mine. But I'd talked to him. So I picked up the phone and argued for him. He runs the function now.
He says he wouldn't hire himself off his own résumé. The system works by exception.
Billions spent on applicant tracking systems. AI platforms on top of those.
Then it's time to decide and someone opens a spreadsheet. Names down the left, one column with no header where somebody typed what they actually think.
That's where the hire gets made.
Traditional hiring flattened craft into a 1-page résumé, but real opportunities come from conversations.
We built spaces in NYC where builders turn talk into co-founders & hires. Now we’re scaling that with AI agents. That’s NightSky.
Double Diamond S2 drops this fall!
AI didn't break hiring. Look at what each part gets paid for.
- Agencies get paid when a role is filled, not filled well.
- LinkedIn earns from recruiters.
- Job boards monetize volume.
- Interviews manage the company's risk.
Add it up and you get a system that runs on pedigree.
A résumé gets about 3 seconds.
In 3 seconds nobody reads it. You scan for shapes you recognize: logos, titles, how long they stayed.
So people write for the scan. The hardest thing they ever built becomes one bullet. 15 years on the reading side. At that volume I'd scan too.
AI is rewriting how startups get funded, built, and defended — valuations on traction that didn't exist six months ago, moats that erode in a single model release.
So we're putting the people who actually make these calls in a room. On the record.
On 8th July we're doing our FIRST live taping of Inside Startup Funding at our space.
Jake Yormak — Managing Partner & co-founder, Story Ventures (and a former startup attorney).
Samir Bakhru — leader of Orrick's Technology Companies Group, $10B+ in financings advised.
What earns a term sheet right now, what "defensibility" even means when the tech resets every quarter, and the legal traps founders hit early. Plus live Q&A.
Link below.