Compliance by design
Mapped to the rules your regulator checks, in each region.
A working prototype takes a week. Getting it past data protection, operational resilience and compliance can take months. We design every flow against the rules where you operate and hand you the documents that show how.
Germany
66%
of companies that use AI name data-protection requirements as an obstacle, the most-cited one.
Bitkom, Sep 2026 · 603 companies with 20+ staff
United Kingdom
No. 1
regulatory constraint on AI in financial services is data protection and privacy.
Bank of England and FCA, Nov 2024 · 118 firms
Banks
67%
of banks name data privacy and security concerns as a top challenge in adopting agentic AI.
EY-Parthenon, Mar 2025 · 100 retail and commercial banks, mostly US
The 3 questions reviewers ask first
We answer them before you're asked.
- 01
Where does our data go?
Nowhere new. Agents run in your cloud account or on your servers, with models in your region or on-premises.
- 02
How does the contract work?
Terms mapped to DORA, UK outsourcing rules or US third-party guidance, with sub-contractors, audit rights and a tested exit plan.
- 03
Who makes the decision?
A named person. Agents prepare and suggest; approvals, overrides and sources are logged in your systems.
One product, any rulebook
Pick your region.
Operating somewhere else? We map every flow to your local rules the same way. Your data protection, risk and compliance teams confirm it, and you stay the regulated entity.
GDPR
A lawful basis for every flow, only the data it needs, and a person behind every significant decision.
EU AI Act
Each agent classified by risk, with transparency, human oversight and logging built in.
DORA
ICT third-party register entry, contract terms, sub-contractors with locations, audit rights and an exit plan.
AMLD and AMLR
Analysts decide (AMLR Art. 76 from July 2027), no tipping-off (Art. 73), AML data never used for sales.
PSD2 and consumer law
Strong authentication before account data, complaint deadlines, a person on request in customer chats.
EBA Guidelines
Outsourcing expectations; explainable, traceable models in loan origination (EBA/GL/2020/06).
Insurance, credit and collections
Demands-and-needs tests under the IDD, forbearance before enforcement under CCD2 (from 20 November 2026), fair treatment by credit servicers.
Germany adds
- BaFin guidance on ICT risks of AI (December 2025) mapped for each agent
- Scoring of individuals kept out of the standard setup (§ 37a BDSG, from 20 November 2026)
- Works agreement template for tools your staff use (works council co-determination)
UK GDPR and DPA 2018
Lawful basis, DPIA and safeguards for automated decisions.
Outsourcing and resilience
FCA SYSC 8 and PRA SS2/21 outsourcing expectations reflected in the contract and exit plan.
Financial crime
MLR 2017 controls, analysts decide, and no tipping-off (POCA s. 333A).
Sanctions
Freezing and reporting decisions under SAMLA 2018 and OFSI guidance stay with your staff.
PSRs 2017, DISP and Consumer Duty
Complaint deadlines met, and customers can always reach a person.
AI guidance
Built to the FCA and PRA principles-based approach to AI and the ICO guidance on AI and data protection.
ICOBS and CONC 7
Claims handled promptly and fairly; forbearance for customers in arrears, with vulnerable customers routed to people.
GLBA
Customer data stays in your environment under your safeguards programme.
BSA/AML and OFAC
Investigations prepared by agents, decisions and filings made by your staff.
ECOA and Regulation B
Credit decisions and adverse-action reasons stay with your credit officers; agents prepare the credit file and list the factors behind it.
Third-party risk
Contract terms mapped to the interagency guidance on third-party relationships (2023).
Incident notification
Drafts support the 36-hour computer-security incident notification rule.
State privacy and AI laws
CCPA rules on automated decisions and Colorado's AI law (both from 2027), and state bot-disclosure laws.
Insurance and collections
State AI bulletins for insurers (NAIC model), unfair claims practices laws, FDCPA and Regulation F for third-party collectors.
Your data protection law
Mapped to the law where you operate, such as Swiss nFADP, Brazil's LGPD, Singapore's PDPA or the UAE PDPL.
Your financial regulator
Outsourcing, operational resilience and AI expectations from your central bank or supervisor.
Data residency
Data and models stay in your country when local rules require it, on local cloud or on your own servers.
AML and sanctions
Built around FATF recommendations and your local AML rules. Analysts decide, never the agent.
AI rules as they arrive
Human oversight, transparency and logging designed in, so new AI laws need a mapping, not a rebuild.
The same documentation pack
Process map, impact assessment, approval matrix and incident runbook, adapted to your regulator.
What our agents never do
- Move money or change limits without four-eyes approval
- File a suspicious-activity report or block an account
- Score people on personality, emotions or social media
- Show compliance status in a message to a customer or counterparty
- Share your data or use it to train models for anyone else
What your reviewers receive
- 01Process and data map
- 02Draft impact assessment
- 03Outsourcing classification
- 04Deployment route and model record
- 05Contract terms mapped to your rules
- 06Shadow-mode results
- 07Approval matrix: automatic, needs sign-off, never automated
- 08Incident runbook with EU, UK and US reporting timelines
Bring your reviewers to the first call.
We'll walk them through the mapping for your region and your process. NDA from the first call if you need it.