OpenAI's 7th Acqui-Hire Deletes Its Data May 13. Source the Rejects.
OpenAI's Hiro Finance acqui-hire deletes user data May 13, 2026. Here is how to systematically source the engineers, designers, and contractors left behind.
OpenAI's acqui-hire of Hiro Finance shut the product on April 20 and will permanently delete user data on May 13, 2026. It is OpenAI's seventh known acquisition this year, after Convogo, Torch, Crixet, OpenClaw, Astral, and Promptfoo. Every other recruiter is chasing the founders into OpenAI. The interesting pool is everyone OpenAI did not put on the badge list.
The pattern is now a calendar, not a news cycle
OpenAI closed six acquisitions in Q1 2026 alone, nearly matching its eight deals across all of 2025. The sector split is wide: developer tools (Astral, Promptfoo), healthcare (Torch), media (TBPN), AI agents (OpenClaw), and now consumer fintech (Hiro). The deal mechanics, though, are nearly identical. OpenAI buys the team, not the IP. The product winds down on a published date. Founders join a research or product org. Everyone else gets a severance check, a LinkedIn headline edit, and a quiet 60 days to find the next thing.
That last group is the sourcing pool no one is systematically working.
For Convogo, an OpenAI spokesperson stated plainly that the company was not acquiring Convogo's IP or technology, only the three co-founders. For OpenClaw in February, the deal was structured around a single Austrian developer, Peter Steinberger, leaving every open-source contributor outside the tent by definition. For Hiro, the public LinkedIn page lists 13 employees, but reporting put the team at "approximately ten" joining OpenAI, and neither side confirmed the exact retention number. The delta is the entire point.
Why the rejected pool is a higher-quality signal than the retained one
The reflex is to chase whoever just walked into OpenAI. That is the worst trade in the market right now. Those people just got a liquidity event, a new badge, and a 12 to 24 month vest cliff they will not break for your seed-stage offer.
The engineer who passed Ethan Bloch's hiring bar at a Ribbit and General Catalyst backed startup, but did not get the OpenAI tap, is the opposite trade. Same credentialing signal. Zero golden handcuffs. Actively updating their profile this week.
The reflex is to chase whoever just walked into OpenAI. That is the worst trade in the market right now.
Hiro is the cleanest live case
Hiro Finance had two co-founders, Ethan Bloch and Rushabh Doshi. Bloch previously built Digit, which sold to Oportun for roughly $211M in 2021. That matters for one reason: the team you can find by filtering "current company: Hiro Finance" on LinkedIn is a tiny slice of the actual talent surface around this deal.
A direct keyword search for "Hiro Finance" surfaces only one current employee profile clearly tagged to the company. The public page count is 13. Reporting cites approximately ten joining OpenAI. So roughly 90% of the Hiro-adjacent population is invisible to firmographic filters.
The adjacency map for Hiro is not hard to draw on a whiteboard. It is hard to actually query, because it lives across Digit alumni from 2014 through 2021, the Ribbit and General Catalyst and Restive fintech portfolios, contractors who never updated a company field, and designers whose only signal is a Dribbble or Read.cv page that mentions Bloch by name. This is the exact failure mode of LinkedIn Recruiter Boolean: the people you want are described, not tagged.
It is also the specific problem Refolk was built for. You describe the person in plain English ("former Digit engineers who did not follow Bloch to Hiro, or who joined Hiro after 2024 and are not listed as joining OpenAI") and get a ranked shortlist across GitHub, LinkedIn, and the open web. The "current company" filter does not need to be right. The description does.
Five things to actually do this week
1. Build the adjacency map before the deletion date
May 13 is a real forcing function. Former Hiro users, contractors, and partner engineers will be logging in one last time to export data. They are going to post about Hiro this week on LinkedIn. They are going to update headlines. They are emotionally closing the chapter.
That window, the seven days on either side of a data-deletion date, is the highest intent moment to send a first message. Not six months later, when they have already signed at Anthropic or Ramp.
Your map for Hiro specifically:
- Digit alumni 2014 to 2021, filtered for fintech or ML roles
- Ribbit Capital, General Catalyst, and Restive portfolio engineers who left in the last 18 months
- Anyone who lists "Hiro" anywhere in profile text, not just the company field
- GitHub contributors to any repo Bloch or Doshi starred or contributed to in the last three years
- Designers and PMs in the AI personal finance niche who followed Bloch on X or Read.cv
2. Treat every prior OpenAI acqui-hire as a future sourcing event
This is the part most recruiters miss. Acqui-hires have a knowable T+90 and T+365 attrition curve.
refolk prompt: Find engineers and designers who worked at Hiro Finance, Digit, or other Ethan Bloch projects, plus contractors and contributors to those products, who are not currently listed as employees at OpenAI. note: Returns a ranked shortlist across LinkedIn, GitHub, and personal sites, including people whose only Hiro signal is in profile text or a portfolio reference, not the company field. slug: 0rqwg3ypsx
### 4. Mine the open-source trail for the deals that were one-person on paper
OpenClaw is the cleanest example. The headline was "OpenAI acqui-hires Austrian developer Peter Steinberger." The reality is that OpenClaw was an open-source AI agent project with a public GitHub history. Every PR author, every issue commenter who got promoted to maintainer, every contributor who shipped a non-trivial feature is a named, public, sourceable lead.
The same applies to Promptfoo, Astral, and any other dev tools acqui-hire. If OpenAI is not buying the IP, the IP is still sitting on GitHub with full commit attribution. That is a sourcing surface, not a footnote. It is also where Refolk's GitHub coverage genuinely earns its keep, because contributor graphs do not map cleanly onto LinkedIn company fields, and Boolean search will not find "person who shipped three meaningful PRs to OpenClaw's agent runtime in 2025."
### 5. Read the IP language for the structural tell
When an OpenAI press cycle explicitly notes the company is not acquiring the IP or the technology (Convogo, Hiro, OpenClaw), that is the tell. The product team is structurally surplus. Designers, PMs, support engineers, infra, QA, customer success. None of them are the asset. They get severance. They do not get the OpenAI badge.
If you see "OpenAI did not acquire the technology" in the lede, your sourcing list for that company should include every non-founder role, every contractor, and every former employee from the last 24 months. Not just the headcount listed on LinkedIn the day of the announcement.
## The Windsurf precedent is the upper bound
If Hiro feels too small to bother with, look at what happened with Windsurf. A nearly $3B Google deal that left behind everyone Google did not hire. Cognition then laid off 30 Windsurf team members and offered buyouts to the remaining 200. That is a known, named, geographically clustered pool of 200+ AI engineers who are explicitly "deal collateral."
The Windsurf situation is the same template as Hiro, just at 20x scale. Most recruiters are still sourcing from the Windsurf cap table page, not from the Cognition buyout list. The Cognition buyout list is where the actual hiring is happening.
```refolk-not-used
What this looks like as a repeatable system
Build three lists and refresh them quarterly.
List A: Live windows. Any OpenAI, Meta, Google, or Microsoft acqui-hire that wound down in the last 90 days. For each, the founder names, the company alumni, the adjacent investor portfolio, and the open-source trail. Hiro, OpenClaw, Convogo all sit on List A right now.
List B: T+365 alarms. Every acqui-hire from the prior 9 to 15 months, indexed by close date. Reach out to the retained founders 30 days before their first vest cliff. Reach out to the retained team 60 days before their one-year anniversary.
List C: Reject pools. The non-retained engineers, designers, contractors, and contributors from every deal on List A and List B, scored by recency of profile change.
You can run all three by hand in a spreadsheet. It will take you about a day per deal and most of a recruiter's bandwidth to maintain. Or you can describe each list in one sentence to Refolk and have the ranked output update itself as people move. Either way, the calendar is the same. The only question is who is working it before the Hiro people sign somewhere else.
FAQ
How many people are actually sourceable from the Hiro Finance acqui-hire?
Public LinkedIn lists 13 employees, reporting cites approximately ten joining OpenAI, and direct keyword search surfaces only one profile clearly tagged to the company. The real sourceable pool, once you add Digit alumni, Ribbit and General Catalyst portfolio overlap, AI fintech contractors, designers, and partner engineers, is significantly larger than the headline headcount. Plan for adjacency, not roster.
When is the right moment to reach out to OpenAI acqui-hire candidates?
Two windows. The first is the 14 days around a public shutdown or data-deletion date (May 13 for Hiro), when non-retained team members are actively closing the chapter and updating profiles. The second is the 30 to 60 days before the one-year anniversary of close, when retained founders and engineers hit their highest statistical attrition risk (33% leave within year one per MIT Sloan and Wharton research).
Is OpenAI's acqui-hire pace likely to continue in 2026?
OpenAI completed six acquisitions in Q1 2026, nearly matching its full-year 2025 total of eight. Meta, Google, and Microsoft are running the same playbook, often labeled reverse acqui-hires, hiring star talent and licensing technology while discarding the rest. Treat the M&A calendar as a recurring sourcing calendar, not a string of one-off news events.
What is the fastest way to build a list of non-retained engineers from a specific acqui-hire?
Start with the founders' prior companies (for Hiro, that means Digit's 2014 to 2021 alumni), then layer in the acquired company's investor portfolio for adjacent moves, then the open-source trail if the product had any public code. Skip "current company" filters entirely; most of the pool will not have updated yet, or will have only ever appeared as a contractor. Describing the pool in plain English to a sourcing tool like Refolk is usually faster than trying to translate the same description into Boolean.