The short version
- Sourcing platforms build a profile of you from public data. SeekOut searches across more than a billion profiles assembled from LinkedIn, GitHub, academic publications and patents.
- For engineers it goes further: SeekOut rates public GitHub activity with a one-to-five-star Coder Score and attaches it to the profile.
- You are not asked, and there is no application. The profile exists whether or not you have applied anywhere.
- In January 2026 two job applicants filed a class action against Eightfold AI alleging it acts as a consumer reporting agency under the Fair Credit Reporting Act without following the statute's disclosure, consent and dispute rules. Eightfold has moved to dismiss. The allegations are unproven, and as of September 2026 the court had not ruled.
- The practical consequence: your public footprint is screening material, and the parts of it you never think about are the parts you do not control.
Almost everything written about getting hired assumes the process starts when you apply. For a large and growing share of roles it does not. It starts earlier, without your involvement, in a system that has already built a record of you from whatever it could find.
This is sourcing rather than screening after you apply, and it is a different industry with different products, different vendors, and almost no candidate-facing documentation. It is worth understanding, because it is the part of hiring where you have the least visibility and, unusually, some genuine control.
What these platforms assemble
SeekOut, Eightfold and Beamery are the names that come up most in enterprise recruiting. They do broadly the same thing: aggregate public and licensed data into searchable candidate profiles, then let a recruiter query that pool directly rather than wait for applications.
The scale is the part worth sitting with. SeekOut says it searches over 1 billion candidate profiles drawn from LinkedIn, GitHub, academic publications, patents and professional databases, alongside the employer's own applicant database. Eightfold's own materials, as quoted in the lawsuit covered below, claim profiles of more than 1 billion people. Beamery's platform includes a talent CRM that customers use to nurture passive candidates over time, and it collects profile updates from individuals or their public profiles.
For technical roles the enrichment gets specific. SeekOut rates GitHub users with a Coder Score of one to five stars, overall and for each programming language, where five stars marks the top 2% of GitHub members. Nobody consents to that score. Nobody is shown it. It is derived from repositories made public for entirely unrelated reasons, sometimes years earlier.
SeekOut's documentation says the score rewards code contributed to projects with many contributors and code that other developers fork often, and gives very few points for solo projects. It also notes that developers with lower scores may still be very good but typically do little open-source work. So a low score can reflect little public open-source work rather than low ability.
The question now in front of a court
In January 2026, applicants Erin Kistler and Sruti Bhaumik filed a putative class action against Eightfold AI in Contra Costa County Superior Court in California. The case was removed to federal court in March 2026 and is now Kistler v. Eightfold AI Inc. in the Northern District of California. The complaint alleges that the platform functions as a consumer reporting agency under the Fair Credit Reporting Act, and as an investigative consumer reporting agency under California's Investigative Consumer Reporting Agencies Act, without the disclosures, consent and dispute mechanisms those statutes would require.
The specific allegation is that Eightfold collects personal data from third-party sources of varying accuracy and runs it through a proprietary large language model to produce a likelihood-of-success score from 0 to 5, which employers use to rank applicants. The complaint quotes Eightfold's own materials describing more than 1.5 billion global data points and analysis of public sources including LinkedIn and GitHub. The plaintiffs argue that such a score is a consumer report in substance, whatever it is called.
“If an algorithmic score about a person, assembled from third-party data, is a consumer report, an entire category of hiring software has been operating without the obligations that come with one.”
It should be stated plainly that these are allegations. Eightfold has moved to dismiss, arguing that it is not a consumer reporting agency and that its software analyses information candidates provide for the employer's internal use. In July 2026 the court vacated the scheduled hearing and said it may rule on the papers; when we last checked the docket, in September 2026, it had not. Nothing has been proven, and Eightfold has not been found to have violated anything.
The legal question is not new. In October 2024 the Consumer Financial Protection Bureau issued a circular stating that an entity which collects consumer data to train an algorithm producing scores about workers for employers could be a consumer reporting agency, and the complaint cites it. The CFPB withdrew that circular in May 2025 along with many other guidance documents, and Eightfold's motion leans on the withdrawal. Several employment law practices have still flagged the case as a potential turning point for the sector.
For a candidate the relevant detail is the one the complaint takes for granted: that scoring from aggregated third-party data is standard practice, and that you have no FCRA-style right to see or dispute what the score says. Applicant tracking systems score too, but vendors describe that scoring as working from the application: Greenhouse says Talent Matching compares your resume against the job's calibration, and Workday describes HiredScore candidate grading as matching resumes against the skills and qualifications in the job description. A sourcing profile is built from material you never submitted.
Check what your resume actually says about you
Optimize your application materials in under 2 minutes.
Worth being direct about the limits here: nothing you do to a resume changes a sourcing profile. Those are separate systems. The resume governs the applications you submit.
What you can actually control
Less than you would like, but more than nothing, and the available actions are cheap.
- Treat LinkedIn as the primary record, because both SeekOut and Eightfold list it as a source and it is the one you can edit directly. Stale titles and missing dates propagate everywhere.
- Keep dates and tenure consistent between LinkedIn and your resume. Aggregated profiles surface discrepancies that a human reader would never have assembled.
- Look at your public GitHub the way a scoring system would. SeekOut's score favours code contributed to projects with many contributors and code other people fork, and gives very few points for solo projects. Abandoned repositories and tutorial forks are contribution history too.
- Fill in the fields that get indexed: headline, location, skills. Empty fields do not read as neutral, they read as absent. Name skills the standard way, since recruiter searches match terms rather than intentions.
- Keep publications and patents attributed correctly, since these are pulled by name and misattribution is common.
- Use the privacy rights you have. SeekOut's candidate privacy policy says that, depending on where you live, you may be able to ask what it has collected about you and request correction or deletion, through its Privacy Choices Portal or by email.
The mental model that helps: your public professional footprint is not a portfolio you present, it is a dataset someone else queries. Those are different things, and only one of them rewards being tidy.
