A critical assessment of the $4M pre-seed building a video-first, AI-interview hiring platform — “TikTok meets LinkedIn” — led by Luminar Ventures with the Candy Crush co-founders as angels. The short-form-video wedge that makes Fika distinctive is also its single largest liability under the EU AI Act.
Unclear, and unaddressed. AI candidate-screening is “high-risk” under the EU AI Act, and emotion recognition in interviews is prohibited as of February 2025. Fika’s “humans make the final decision” disclaimer does not obviously exempt an AI that ranks and surfaces matches.
It is baked in. Putting race, age, gender, appearance, and accent in front of an employer before qualifications is the product, not a side effect — and TechCrunch flagged exactly this in the launch coverage. No mitigation has been described.
Only if candidate-side virality builds a two-sided network rivals can’t copy. Mercor, Alex, and Maki are far better capitalized, so Fika’s survival depends on owning the consumer experience before regulation or a larger player neutralizes its one differentiator.
Key Finding: Fika Jobs has raised a credible $4M pre-seed from strong consumer-pedigree backers to build a genuinely differentiated “TikTok-meets-LinkedIn” hiring product — but its single biggest design choice, forcing video to the front of every application, is simultaneously its competitive wedge and its largest legal liability. The EU classifies its category as high-risk and bans the exact emotion and trait inference a video product invites, and the company has published nothing on how it navigates that.
The most important question about Fika isn’t whether candidates like short video — it’s whether a Stockholm-based AI that screens and ranks people is allowed to operate the way it’s designed to.
AI reviews background, generates personalized questions. Profile + CV data ingested.
~10-minute interview run by Google Gemini. Captures audio, video, transcripts.
Responses cut into short, TikTok-style clips — the differentiator and the exposure.
Ranks candidates to roles on “9+ parameters.” Ranking is itself regulated activity.
Race, age, gender, accent, appearance visible before qualifications. Bias door opens.
One-click apply; 10% first-year-salary fee on a placement — the only revenue event.
A “human clicks the final button” framing has not historically exempted screening tools from high-risk classification. Fika’s privacy policy admits the AI is used to “assess profile relevance, surface matches” — and ranking and surfacing candidates is the regulated act, not just the final hiring decision.
Fika’s differentiation is forcing video to the front of every application. But the EU AI Act classifies AI hiring tools as “high-risk” — triggering bias testing, documentation, human oversight, and conformity assessment — and as of February 2025 it prohibits emotion recognition in interviews and video assessments outright. A video-first product built to read personality and “grit” from facial expression and tone is aimed squarely at the banned zone. Nowhere in the launch materials, the website, or the privacy policy is there any mention of EU AI Act conformity, bias auditing, or how Fika avoids the prohibition.
Tier-1 Nordic seed fund — lead investor; credible regional signal
King / Candy Crush co-founders — consumer-virality pedigree that maps to the “TikTok for hiring” thesis
Powers the entire AI interviewer. Fika controls neither its behavior, pricing, nor bias profile
TechCrunch and Fika’s own site name different beta customers; only Rebtel and PlentyLabs overlap
Privacy policy names zero sub-processors despite collecting video/audio of EU job seekers
10% of first-year salary ties revenue to high-salary placements — pulling against neutral matching
The funding event is well-corroborated. The usage numbers are one day old, company-supplied, and inconsistent across sources.
The “alignment” pricing story has an unexamined edge. A 10%-of-first-year-salary model is framed as candidate-friendly, but it makes Fika’s revenue a function of placing candidates into the highest-salary roles — a structural incentive to optimize for closed, expensive placements over best-fit matching.
Seven structural risks that the $4M pre-seed does not resolve.
A Stockholm-based AI that screens and ranks candidates is a textbook “high-risk” system requiring conformity assessment, bias testing, and documented human oversight. A video product that reads tone and expression edges toward the Act’s outright ban on emotion recognition in hiring. Fika has published nothing on compliance.
Exposing race, age, gender, appearance, and accent before qualifications is intrinsic to a video-first model, not a tunable bug. TechCrunch flagged it in the launch coverage itself. This is the structural risk the company has never publicly addressed.
The privacy policy collects video, audio, and transcripts of EU job seekers but names zero sub-processors, sets no retention periods, and only vaguely concedes non-EEA transfers. That is a documentation deficit on the most sensitive data category in the product.
$4M against a $10B Mercor and well-funded Alex and Maki, with a differentiator — short-form video — that is also the product’s biggest legal liability. The wedge can be copied by anyone willing to absorb the same risk.
The entire AI interviewer runs on Google Gemini. Fika controls neither the model’s behavior, pricing, availability, nor bias profile — and inherits any Gemini fairness issue into a regulated hiring decision with no disclosed fallback.
“100+ waitlist / 50+ tested” are unverifiable and the beta-customer list differs between TechCrunch and Fika’s site. There are zero confirmed placements — and the 10% success fee structurally tilts toward expensive roles over best-fit matches.
Fika Jobs is a well-backed, emotionally well-positioned bet on making hiring feel human. But it is launching the most legally exposed possible product — mandatory candidate video — in the most aggressively regulated possible market, and it has said nothing about how it survives that contradiction. The investor signal and UX differentiation are real; the regulatory, bias, and competitive risks are structural and unaddressed. The upside is a genuine consumer category; the downside is that its core mechanic is the thing EU regulators are actively trying to stamp out.
Based entirely on publicly available information, including the TechCrunch announcement of June 23, 2026. Every number is labeled CONFIRMED, DERIVED, or EST in the body; unverifiable company claims are flagged as such.
