PowerToFly found the people. Now AI needs them.

The curious thing about PowerToFly’s new AI business is how much it owes to an old hiring problem. In 2014, Katharine Zaleski and Milena Berry founded a company to connect women with remote and flexible work. Twelve years later, the company is recruiting people to help judge artificial intelligence. The intervening story concerns a remarkably persistent habit: overlooking expertise because it arrives in an inconvenient package.

The résumé had a blind spot

Zaleski’s experience after having her first child helped give the business its purpose. A career could remain valuable even when its owner needed different working arrangements. Berry brought experience as chief technology officer at Avaaz, where she had built a distributed technical team. Between them, they understood both the demand for flexibility and the machinery required to make it practical.

Zaleski later made her change of mind unusually public. Her 2015 essay apologized to working mothers she had judged before becoming one herself. There is something useful about an entrepreneur admitting that she once shared the assumption her business exists to challenge. The customer problem had become personal enough to overturn her own previous convictions.

The interview needed an interview

An early company account describes a three-week assignment to hire women into a media company’s technology team. PowerToFly invited 400 women to a webinar with hiring managers. Afterwards, 141 expressed interest. Technical review and candidate conversations produced a shortlist of 22; seven were hired. The webinar gave candidates a chance to decide whether the employer deserved their attention.

The shortlist exposed another obstacle. Candidates questioned a heavily male interview panel and an interviewer who seemed more interested in appearance than technical ability. PowerToFly relayed the feedback; its account says the hiring manager changed the process. The immediate failure was inside the interview. Supplying more candidates would have left it intact.

Put the engineers on the invitation

That sequence helps explain the company’s place between a recruiting service, an employer storytelling business and a professional community. Its customers need access to qualified people, but also a reason for those people to engage. A job description can list requirements. An engineer discussing the actual project can reveal whether the work is interesting enough to change jobs for.

In PowerToFly’s Salesforce case study, engineering leaders appeared at eight dedicated hiring events. The company reports 57 hires across technical and commercial roles. The arrangement combined events with sourcing, screening and ongoing employer visibility. One attendee came away willing to consider joining Salesforce, an unglamorous but commercially valuable change of mind.

“Now I would seriously consider Salesforce as my next place of employment.”Event attendee · Salesforce case study

Autodesk’s published story describes a similar mismatch: plenty of applications, too few qualified candidates for specialized positions. PowerToFly combined employee-led events, sourcing support and branding content. Its case study reports 30 hires and 85% retention after five years. These are the vendor’s figures, without a comparison group. Still, they describe an outcome more meaningful than a crowded inbox.

An education company walks into the talent pipeline

PowerToFly broadened beyond its original focus on gender and remote work into recruiting, retention and development for underrepresented professionals. The founders’ 2022 funding note acknowledged the limits of their earlier perspective and the need for a more intersectional approach. Their account also described six months of fundraising and repeated rejection. Expansion had a bill, and collecting the money took persistence.

The commercial model puts employers on the buying side: recruiting services, events, branding and talent software. Core job discovery and community resources are free for professionals, while education can carry separate fees. The appeal for an employer is a combination of access and assistance, rather than another list of names to sort alone.

The September 2022 financing announcement put that bill at more than $30 million in combined financing, led by Morgan Stanley Expansion Capital and including Grasshopper Bank debt. It backed learning resources and a broader recruiting-and-retention platform. Earlier, a $6.5 million Series A in 2015 had followed a $1 million seed round. The capital supported successive versions of the business.

In May 2024, PowerToFly acquired Skillcrush, adding technical education to its hiring operation. Skillcrush founder Adda Birnir said the relationship went back to a Romanian designer hired through PowerToFly in 2015. A customer had become part of the company. Training could now feed a talent pipeline, giving the combined business something to offer before a professional was ready to apply.

Now the machine needs a second opinion

The September 30, 2026 announcement schedules an official AI workforce launch for October 14. PowerToFly lists clients including Visa, Deloitte, Zapier and Arize AI. Its proposition applies the existing network to a new task: finding people who can recognize whether an AI answer meets the expectations of a particular industry.

The employer offering has two routes. Companies can hire full-time, contract or project-to-hire professionals. They can also engage experts for human feedback, model evaluation and red teaming, which deliberately probes weaknesses. A clinician, lawyer or linguist brings a different test of correctness from a general reviewer. PowerToFly markets credential verification and continuity across model versions as part of that service.

This puts it beside expert-work networks such as Mercor and AI data providers such as Scale, while its recruiting side overlaps with employer discovery services such as The Muse. PowerToFly’s distinctive asset is the community assembled through inclusive hiring. That history makes its move understandable; it does not by itself establish better model performance. Buyers still need to test quality against their own tasks.

What to borrow, what to try

An employer can borrow the hiring method immediately: let candidates hear from future colleagues, make the work specific, and invite criticism of the interview. It depends on managers who will answer candidly and change a poor process. For an AI project, define acceptable output before commissioning expert feedback; an impressive credential cannot rescue a vague assignment.

Professionals can create a profile, explore jobs and attend virtual events. The candidate site advertises AI project rates typically around $65-130 an hour, depending on expertise and assignment, and says coding is not always required. It is also assembling its first cohort, so an application should not be mistaken for immediate paid work. The practical first step is to describe what you know precisely enough that someone else can put it to use.