How Floats skipped a senior engineer-year of ATS integration work with Unified.to
September 1, 2026
Product Category: AI candidate profile generation for staffing agencies Use Case: Read candidate data from ATS integrations, write profile data and status back Features Highlighted: ATS integration coverage, pass-through architecture, per-integration read and write capability detail Benefits Highlighted: Faster time to market, shorter security review, no per-ATS build cost
Floats turns raw candidate data into client-ready candidate profiles. A staffing agency selects a candidate in its ATS and Floats generates a branded, structured profile the recruiter can send straight to a hiring manager, rather than rebuilding the CV by hand in Google Docs.
The harder part of a submission is not formatting. It is the research: understanding what a candidate actually did, what their previous employers actually do, and whether that maps to the brief. Floats does that research and surfaces the evidence, so recruiters spend their judgement on the match rather than on building the case for it.
"Integrations weren't a nice-to-have for us, they were the entry ticket. If a candidate's data can't get out of the agency's ATS, there's no product. So every deal ran into the same wall: 'we're on Bullhorn / JobAdder / Loxo / Recruit CRM, do you support it?'"
— Scott Anderson, Co-founder, Floats
Building the integration layer in-house was most of an engineer-year
Floats looked at three options: build the integration layer themselves, go direct to the one or two ATS platforms their earliest customers used, or use a unified API.
Going direct was the first one out. It optimizes for the customers you already have and blocks the ones you want.
Building in-house came down to the maths. Every ATS is a separate build with its own auth model, pagination, candidate object shape and rate limits, plus partner-program approval and sandbox access before an engineer can start.
What decided it was the opportunity cost. The team is small, and what makes Floats valuable is the profile generation, not the integration layer underneath it. Spending the first year building an ATS abstraction would have meant pushing out AI capability and losing the window on the part of the product they are differentiated on.
"Every prospect on an ATS we hadn't built yet was a deal we couldn't credibly pursue. We'd have been telling them 'in two quarters,' which in staffing means no."
— Scott Anderson, Co-founder, Floats
On the unified API side, Floats evaluated Merge, which the team describes as the enterprise-mature option in the category. It came down to two things: coverage of the ATS platforms their segment actually runs on, and pricing that works at their stage.
→ The two architectures are broken down in Merge vs. Unified.to: A 2026 Comparison for SaaS and AI Teams
Why Unified.to is the right integration partner
Nothing holds a copy of the candidate database. Reads pass through to the source ATS and transform in memory. The only things that persist on the Unified.to side are the connection ID and minimal operational metadata, encrypted at rest.
For Floats that matters because of who the data belongs to. Candidate records are third-party PII at two removes: Floats handles them on behalf of the agency, and the agency handles them on behalf of the client companies it recruits for. On a sync-and-store model Floats would have inherited a much larger surface area to defend.
"Being able to say that neither we nor our integration layer holds a mirrored copy of their candidate database materially shortens security review."
— Scott Anderson, Co-founder, Floats
Write support documented per integration. Read-only would only get Floats halfway, because profile data and candidate status have to go back into the ATS. Unified.to publishes read and write capability for each integration, so the team knew what they could write to which ATS before they committed to it.
"That sounds mundane until you've been six weeks into an integration and discovered the write path you assumed doesn't exist."
— Scott Anderson, Co-founder, Floats
Breadth they haven't grown into yet. Floats' roadmap runs past ATS into HR and meeting notetakers. Those are the same connection model and the same authorization component rather than a new evaluation, which makes the build-versus-buy call for the next category obvious.
The result was deals they could pursue
When a prospect names an ATS, Floats can say yes. They went from one supported ATS to publishing support for 140+ ATS platforms on their own site, and the incremental cost of adding one is hours rather than a roadmap conversation.
The year Floats did not spend building an ATS abstraction layer went into the product instead. SOLA, their AI recruiting assistant, is in beta now: conversational search across the agency's own ATS data, candidate matching, and research on the candidates it returns.