Refolk launches: an AI agent that finds the exact engineers you need - and writes to them

Recruiting tools have spent the last decade asking the same question in different shapes: which boolean string, which filter combo, which saved search will surface the right person? Refolk, launching out of San Francisco, is betting the question itself is wrong.

Refolk is an AI agent. You describe who you want in a sentence. It goes and finds them.

What it actually does

The product is built around a single search box that accepts plain English. The demo on refolk.ai walks through a query that would take a sourcer half a day to translate into LinkedIn Recruiter syntax:

Senior Rust engineers in San Francisco, ex-FAANG, now at small teams.

Refolk's agent fans out across GitHub, LinkedIn and the open web, ranks candidates against the brief, and returns a shortlist within seconds. Crucially, it does not stop at the list. For each person it surfaces, the agent drafts a personalized outreach email - referencing the specific repo they maintain, the company they helped scale, or the talk they gave last spring. The kind of detail that takes a human recruiter twenty minutes per send and gets done for two of the fifty names on the spreadsheet.

The interface gives operators levers they expect: filter by industry (AI, SaaS, fintech, dev tools, cybersecurity, biotech), city, funding stage, investor. There is also a direct repo-search mode, useful when the search target is "someone shipping production Rust on a real distributed system" rather than a job title.

Why now

Two things changed in the last 18 months that make Refolk's bet plausible where similar pitches would have stalled in 2022.

The first is that LLM-driven agents have crossed the threshold where they can read a GitHub profile, a LinkedIn page, and a company blog, and produce a coherent paragraph that does not feel like spam. Personalization at scale has gone from a marketing slogan to a primitive you can rent by the token.

The second is that hiring teams are smaller and more selective than they were two years ago. The 2024-2026 contraction in tech hiring has not made recruiters disappear - it has made them more accountable for response rate and quality of pipeline. A tool that gets a 20% reply rate where a generic blast gets 2% pays for itself in a single hire.

The market it is walking into

The talent intelligence and outreach space is not empty. LinkedIn Recruiter remains the default. Newer entrants like Gem, hireEZ, SeekOut and Juicebox have each carved a slice. The gap Refolk is targeting is the one between "search tool that gives you a list" and "writes the email for you" - and doing both in the same agentic loop, without the operator stitching the workflow together across three tabs.

The product is also positioned for a use case that adjacent tools have not handled well: business development and investor sourcing. The same engine that finds Rust engineers can find founders building in dev tools, or angels who have written checks into early-stage cybersecurity. Refolk is described on the site as useful for recruiters, talent acquisition teams, and BD professionals - which is consistent with how the underlying agent is built.

What is unclear

Pricing is not on the launch page. The site lists hello@refolk.ai as the contact, suggesting an early-access or hand-rolled onboarding process while the team calibrates the right tier structure. Likely candidates are seat-based for recruiters and usage-based for high-volume sourcing teams, but Refolk has not committed publicly.

Two questions matter for buyers evaluating the product.

First, data freshness. A recruiting agent is only as good as how recently it indexed the candidate's GitHub activity and LinkedIn role. Stale data produces outreach that opens with "congrats on your new role at Company X" when the candidate left X six months ago. Refolk has not yet detailed how often it re-crawls.

Second, defensibility against the platforms. LinkedIn has a long history of pushing back on third-party scrapers via litigation and rate limiting. GitHub is more permissive but rate-limits aggressive crawlers. The legal and engineering posture Refolk has taken on this matters for whether the product is durable or whether it gets quietly throttled into uselessness.

Bottom line

Refolk is one of the cleanest examples to date of an agent product that collapses three tools - sourcing, enrichment, and outreach drafting - into a single English-language interface. Whether it becomes a category leader or a feature LinkedIn copies depends on execution and on the data-access fight that every web-scale recruiter eventually picks. But the demo is real, the wedge is sharp, and the timing is right.

You can try it at refolk.ai.