Equals Group: onboarding due diligence, run by AI agents.
Fifty One Degrees designed, built and delivered a multi-agent AI system for Equals Group, a UK fintech providing payments and currency services to consumers and businesses. The system runs eight due diligence checks on every SME onboarding case and presents its findings inside the tools analysts already use, with a human reviewing every decision.
Onboarding checks that scaled only with headcount
Equals Group provides payments, currency and account services to consumers and businesses across the UK and Europe. Each new SME client needs know-your-business due diligence before onboarding: incorporation records, address verification, website checks, document review, adverse media screening. Every check was manual, every case took analyst time, and demand was growing faster than it made sense to grow the team. Equals needed the routine checks automated without compromising the quality or compliance of a regulated onboarding process.
One specialised agent per check
Eight automated due diligence checks
Address verification, incorporation verification against Companies House, website domain analysis, website content analysis, website visual analysis, LinkedIn profile analysis, adverse media screening and document analysis. Each check is handled by its own specialised AI agent.
Verdicts with reasoning, not black boxes
Every check returns pass, fail or uncertain together with an explanation of why, so analysts can see the evidence behind each verdict rather than trusting a score.
Inside the tools analysts already use
Results surface directly in the onboarding team's existing case interface via a REST API. No new system to learn, no swivel-chairing between screens.
A human on every decision
The agent does the legwork; people make the calls. 100% of cases are reviewed by a human before any onboarding decision, by design.
Built for a regulated fintech
Scoped at the analysts' desks
Fifty One Degrees shadowed Equals' onboarding analysts through real cases before writing the scope, so the agents automate the checks as they are actually done.
A multi-agent Python system
Built in Python with the PydanticAI agent framework and FastAPI, one agent per check, delivered as a Docker image with observability built in.
Google Gemini, no training on client data
Reasoning runs on Google Gemini under a strict no-training policy, with specialist tooling for company registry lookups, web analysis and document fraud detection.
Privacy by architecture
Customer personal data is processed transiently and never persisted by the system, with API-key authentication and encrypted transport throughout.
Proof first, then a careful release
The system went from kickoff in February 2025 to full release in around seven months, through a release plan built for a regulated business: prove it, run it in parallel, release it in stages. The engagement also broadened into AI strategy work across the wider group.
A working proof of concept demonstrated on real onboarding cases within weeks of kickoff.
Scope each of the eight checks with the analysts who run them, including on-site shadowing.
Build the full multi-agent system and integrate it into Equals' onboarding interface.
Run the agent alongside the existing manual process, comparing verdicts case by case.
Release to a controlled slice of live cases over three weeks, with every decision human-reviewed.
Hand over to business as usual, with the architecture designed to extend to new markets and deeper checks.
What the system delivers
Designed to move up the funnel
The architecture was built to extend: enhanced due diligence for higher-risk cases, new markets beyond the UK and EU, and a risk-appetite agent at the top of the funnel that pre-screens prospects before sales effort is spent. Equals' teams projected a 50% productivity gain for onboarding and a further 30% for sales once the agent moves to the top of the funnel.
Fifty One Degrees delivered Equals Group a multi-agent AI system that runs eight due diligence checks on every SME onboarding case, each returning a pass, fail or uncertain verdict with its reasoning attached, inside the interface analysts already use.Fifty One Degrees x Equals Group
What did Fifty One Degrees build for Equals Group?
A multi-agent AI system that automates eight know-your-business due diligence checks on SME client onboarding: address verification, Companies House incorporation checks, website domain, content and visual analysis, LinkedIn analysis, adverse media screening and document analysis. Each check returns pass, fail or uncertain with an explanation.
Does the AI make onboarding decisions?
No. The agents do the evidence-gathering and checking; 100% of cases are reviewed by a human before any onboarding decision. Every verdict comes with its reasoning attached so analysts can verify it.
How was the system released safely?
Through a phased plan built for a regulated fintech: proof of concept, a parallel run alongside the existing manual process, a three-week controlled release on live cases, then full release. Customer personal data is processed transiently and never persisted.
Can Fifty One Degrees automate onboarding or KYB checks for my business?
Yes. Fifty One Degrees builds custom AI agents around a firm's real compliance obligations and systems. Book a Discovery Call and we will map which checks an agent could take on first.
Want onboarding that scales without headcount?
Book a 30-minute Discovery Call and we'll map which of your due diligence and onboarding checks an AI agent could run, with humans kept on every decision.