- Home
- Industries
AI agents by industry: regulated sectors, each with its own line.
The AI Act lists some uses as high-risk sector by sector, and GDPR treats some data as special. We start where an agent helps most and the risk is easiest to manage, and we know where the line is.
Industries we serve
Banks and financial services
KYC files, AML alert preparation and customer questions, kept clear of credit scoring and contracted on DORA’s terms.
AI Act Annex III 5(b)AI Act Art. 6(3), 6(4)Insurance
Claims intake, policy answers and complaint files, with a handler deciding every payout. Life and health pricing stays off-limits.
AI Act Annex III 5(c)AI Act Art. 27Consumer lending
Loan support, complaints, documents and payment plans for lenders and lending platforms. No agent approves, prices or scores a loan.
CCD2 Art. 18(3)CCD2 Art. 18(9)Health and life sciences
Referral intake, letters for sign-off and regulatory drafting, on the administrative side of the line. No diagnosis, no triage.
GDPR Art. 9GDPR Art. 35Legal and professional services
Contract review against your playbook, due diligence and intake, from your documents. Lawyers stay the ones who advise.
AI Act Annex III 8GDPR Art. 28Public sector
Citizen questions, casework preparation and tender checks, with the AI Act’s public-body duties built in. No benefit decisions.
AI Act Annex III 5(a)AI Act Art. 27Industry and manufacturing
Quality records, supplier documents and technical files, drafted for an engineer to sign. Never a safety function.
NIS2 Annex IIGDPR Art. 5(1)(c)Also: energy and utilities, logistics, telecoms and other companies with regulated processes. The method is the same; the scoping call finds where the Act applies. For a first view on your own, try the AI Act risk-class screener.
Four jobs, step by step, with the person inside the flow.
Each run shows what the agent does, what it flags, where it stops for a person, and what stays on record, with the article each step answers to.
Insurance claim intake
A home contents claim arrives by email. The agent prepares it; a claims handler decides the money.
- Sector
- Banking and insurance
- Risk class
- Not listed in Annex III, reasons kept
- Who decides
- A claims handler
- Claim receivedEmail from a policyholder with 3 attachments: photos, a receipt, a police report.Done
- Only the claim file readThe agent reads this email, its attachments and the policy wording; nothing else in the case system.GDPR Art. 5(1)(c)Done
- Checked against the policyTheft from the home is covered; €150 excess. Source: policy wording, section 4.2.Done
- Missing receiptNo receipt for the laptop. A request to the policyholder is drafted, opening with a note that an AI wrote it.AI Act Art. 50Flagged
- Payout proposed€1,240 after the excess, once the receipt arrives, with the reasoning and its sources.Drafted
- Waiting for a claims handlerNothing is paid or refused until a named person decides.Approve €1,240ChangeRefuseAI Act Art. 14GDPR Art. 22Waiting for a person
- Request, reply and payout sentOnly after the handler’s decision.Locked
- Run recordedInput, sources, tools, model, cost (€0.31, sample) and who decided.AI Act Art. 12On record
KYC onboarding check
A new business customer’s file arrives. The agent checks it against the bank’s checklist; an analyst makes the onboarding decision.
- Sector
- Banking and insurance
- Risk class
- Not credit scoring, so not Annex III 5(b); reasons kept
- Who decides
- A compliance analyst
- File receivedID document, proof of address and a company register extract.Done
- Only the fields the checklist needsName, date of birth, address, register number. The rest of the ID image is not kept in the prompt.GDPR Art. 5(1)(c)Done
- Checked against the checklist6 of 7 items match, each with the page it was read from.Done
- Gap foundThe proof of address is five months old; the checklist asks for three.Flagged
- Waiting for a compliance analystThe agent never accepts or refuses a customer.Accept the fileAsk the customerRefuseAI Act Art. 14GDPR Art. 22Waiting for a person
- Customer told the outcomeOnly after the analyst’s decision.Locked
- Run recordedEach check with its source page, the flag, and who decided.AI Act Art. 12On record
Patient letter for sign-off
After an outpatient visit, the agent drafts the follow-up letter in plain words from the clinic’s note and templates. A doctor signs it or not.
- Sector
- Health and life sciences
- Risk class
- Administrative: no diagnosis, no treatment advice
- Who decides
- The treating doctor
- Request receivedA follow-up letter for an outpatient visit, in German.Done
- Only what the letter needsThe visit note’s summary and next appointment; nothing from the wider record.GDPR Art. 5(1)(c)GDPR Art. 35Done
- Letter draftedFrom the clinic’s approved template, each sentence traced to the note.Drafted
- One sentence held backThe note mentions a possible cause. The draft leaves it out and marks it for the doctor.Flagged
- Waiting for the doctorNo letter leaves the clinic unsigned.Sign and sendEditDiscardAI Act Art. 14Waiting for a person
- Letter sentOnly after the doctor signs.Locked
- Run recordedThe draft, the doctor’s changes and the signed version.AI Act Art. 12On record
Contract clause review
A supplier sends its standard agreement. The agent compares every clause with the firm’s playbook; a lawyer accepts or rejects each finding.
- Risk class
- Not listed in Annex III, reasons kept
- Who decides
- The responsible lawyer
- Agreement receivedA 24-page supplier agreement, as a PDF.Done
- Compared with the playbook38 clauses against 12 playbook positions.Done
- 3 deviations foundLiability cap above the playbook; no processor terms for personal data; a 24-month auto-renewal. Each with its source passage.GDPR Art. 28Flagged
- Fallback wording proposedFrom the playbook only, one proposal per deviation.Drafted
- Waiting for the lawyerEvery finding is accepted or rejected by a person; the agent gives no advice.AcceptRejectRewriteAI Act Art. 14Waiting for a person
- Mark-up sent to the supplierOnly after the lawyer’s decisions.Locked
- Run recordedFindings, sources, the lawyer’s decision on each, and the model used.AI Act Art. 12On record
Scripted examples on sample data, not client work. Every name, amount and document above is made up. The steps are how we design these agents on the platform: the agent prepares, a named person decides, and the run is kept.
Bring one process. Leave with a risk class, a pass bar and a plan.
A 45-minute scoping call with the people who would build it. You get a first view of the AI Act risk class, the controls it needs, and whether an agent is the right tool at all.