In insurance, artificial intelligence does not stumble on technology. It stumbles on two constraints the sector knows well. Incoming documents are imperfect by nature, and the regulatory frame is tightening fast. So the real question is not whether AI works, but where to place it so that it holds up in production.
Tandem works with brokers, mutual insurers and insurance groups on exactly that question. This article gives the answer we give on assignments. Four use cases can ship today without falling under the high-risk regime of the EU AI Act. Three others look more profitable on paper, but they sit in a scope that comes under supervision on 2 December 2027.
Where does AI actually stand in French insurance?
AI is the leading transformation driver for insurance jobs. That is the finding of the forward-looking barometer of the Observatory for Insurance Careers, published by France Assureurs in June 2026. The observatory places usage on three grounds, data analysis, fraud detection and task automation. It adds a nuance teams remember, AI reshapes the content of jobs more than it removes them.
Two announcements from the summer of 2026 show the sector moving on two distinct fronts. In July 2026 AXA announced a large-scale rollout of Microsoft 365 Copilot to its staff, three years after launching its internal Secure GPT service. That move is about internal productivity.
MAIF went the other way. In June 2026 the mutual insurer opened an app inside ChatGPT. It rates a home's exposure to climate risk from an address, and estimates the price of bicycle cover (La Tribune de l'Assurance). That is not productivity, it is distribution. The insurer meets the prospect where the question is asked.
Which AI use cases should insurance open first?
The ranking that matters crosses two axes, operational gain and regulatory exposure. Four use cases combine real gain with low exposure. Those are the ones to open first, because they produce value without triggering the heavy obligations of the high-risk regime.
- Inbound document flow. Cancellations, mandates, supporting papers, claim documents. AI reads, structures and routes. It is the largest and least risky pool of time.
- Contract and cover summaries. An advisor gets in seconds what used to sit inside thirty pages of terms.
- Phone reception. A voice agent qualifies the call, answers common questions and hands the rest to a person.
- Claims and regulatory monitoring. Continuous watch on changes, with digests sent to the teams concerned.
Fraud detection deserves a separate mention. The French insurance fraud agency ALFA recorded 946 million euros of fraud detected by its members in 2025. Fire, accident and miscellaneous risk lines account for 627 million of that on their own. The topic is large, and the EU regulation explicitly excludes financial fraud detection from the high-risk regime. The constraint here is not regulatory, it is technical.
This hierarchy is not specific to insurance. We apply the same grid in accounting firms, where inbound document flow also comes first.
See our analysis of seven AI use cases for accountants, which shares the same legal deadline and supporting document constraints.
How Santiane automated part of its support
Santiane is a health insurance brokerage group, bringing together the Santiane, Neoliane and Julia Mutuelle brands. Handling inbound requests was entirely manual. Expert staff spent their days on repetitive tasks, with regulatory deadlines to meet on every file.
The difficulty was not volume. It came from the nature of the problem. Documents arrive scanned, with typos and heterogeneous formats. Business rules must apply without interpretation. And processing health data ruled out any third-party API hosted outside the European Union.
The workflow runs on the group's own infrastructure. Each request triggers two extractions in parallel, one tuned for scans and one for text files, then the two streams merge. No value is ever inferred. A confidence score is computed and uncertain fields are flagged rather than filled with a guess.
No health data left the controlled perimeter, and every file with something at stake stays validated by a person. The full project is published in the Santiane case study. It is the component we reuse most often in brokerage, because inbound flow is the same problem everywhere.
What the AI Act and the ACPR change for an insurer
This is the part AI projects in insurance underestimate most. Regulation (EU) 2024/1689 of 13 June 2024 classes some systems as high risk, with reinforced obligations on documentation, bias management and human oversight. Two insurance uses sit in that scope.
- Risk assessment and pricing of natural persons in life and health insurance.
- Creditworthiness evaluation of natural persons and credit scoring, excluding systems used to detect financial fraud.
The ACPR will be the market surveillance authority on these use cases. It takes up that role on 2 December 2027, and it is not waiting for the deadline to move. In July 2026 it published a discussion paper on algorithmic fairness in the financial sector. The text is open to public consultation until 30 September 2026. In parallel, the authority is building an assessment methodology for the financial sector's AI systems.
One technical point in that paper deserves the attention of any team building a pricing model. Advanced systems are increasingly able to reconstruct sensitive personal characteristics, even when those variables have been removed from the modelling. Dropping the field is therefore no longer enough to demonstrate the absence of discriminatory treatment.
The voice agent, brokerage's most underrated use case
A brokerage loses business on calls nobody picks up. An insurer loses satisfaction on claim calls left on hold. A voice agent handles both, and it goes live in a matter of weeks.
The mechanics come down to four components. A telecom operator carries the line. A transcription turns voice into text. A model generates the answer. Speech synthesis makes it audible. Each component is configured separately, which keeps cost and voice under control.
The trade-off that decides a voice agent's success is not about technology, it is about scope. I have found that an agent loaded with too many scenarios gets lost among them. Three or four well-written resolution paths, plus a human transfer for everything else, beat any attempt to cover every case. In my own tests, handled calls last under a minute on average.
Outbound calls are the forgotten side of the topic, and the one that speaks loudest to insurance. An annual renewal to confirm, an unpaid premium to chase, a prospect who just filled in a form and needs a call back within the minute. The list starts in a spreadsheet, the agent calls, leaves voicemail, sends an SMS and updates the status until it gets an answer.
Steering runs on transcripts. Every call leaves a full text, a duration and a category, which makes it possible to alert teams on unexpected categories. For the complete method, see our AI voice agent guide.
Should you build your AI agent or buy it?
This is the question that stalls the most executive committees in the sector. In 2023, everyone wanted to build. Since then off-the-shelf solutions have multiplied, technical debt has become frightening and the return on internal projects often disappoints.
I laid out this trade-off in a LinkedIn post on the build or buy choice for AI agents. The rule I take from it is simple. Buy when your processes look like everyone else's, build when the topic is strategic. Buy to move fast, build to differentiate.
Insurance has one feature that shifts the build or buy dial. As soon as health data enters the flow, the hosting constraint rules out part of the off-the-shelf market straight away. The Santiane project is exactly that case. The decision to build came from an obligation on the data, not from a technical preference.
| Situation | Our recommendation |
|---|---|
| Phone reception, appointment booking | Buy a platform, the process is standard |
| Contract summaries for advisors | Buy, then tune the instructions to the portfolio |
| Document flow carrying health data | Build in a controlled environment |
| Routing under a legal deadline | Build the rules module, whatever you chose upstream |
| Portfolio-wide fraud detection | Hybridise, a bought base plus in-house rules |
On tooling, we build these workflows on n8n, which installs on the client's infrastructure when sovereignty requires it. See our n8n guide and our n8n agency offer.
How to choose an AI agency specialised in insurance?
The AI provider market is crowded, and most of them have never seen a cancellation letter or a delegation mandate. Five questions sort them quickly.
- Do they understand the insurance chain? A provider who confuses insurer, delegated administrator and broker will make routing errors.
- Can they deploy in a sovereign environment? Ask for a self-hosted workflow reference, not an intention.
- Where do they put the regulatory decision? If the answer is "in the prompt", move on to the next one.
- What does the system do when it is unsure? An explicit handover to a human is a sign of seriousness, an always-confident answer is a risk.
- When does the first deliverable land? On our assignments, the first useful scope runs within weeks, not in six months.
Tandem works across the whole chain, from claim declaration to broker reporting. The use cases by role are on our AI agency for insurance and brokerage page. The scoping approach is described on our AI audit page.
Which AI use case should insurance start with?
Start with the inbound document flow. The gain is immediate, the measurement is simple, and the topic does not fall under the high-risk regime of the EU regulation. You free expert resources within weeks, and along the way you build the component everything else depends on, reliable extraction and explainable routing.
Keep life and health pricing for later, and prepare it seriously. The 2 December 2027 deadline is coming, the ACPR consultation is open, and an undocumented pricing model will be a liability. In between, phone reception and contract summaries make excellent second projects.
One last rule, valid in every case. If your system cannot say that it is unsure, it is not ready for insurance.
