80% of policyholders say they do not understand their benefits tables. Yet the majority of their questions can be resolved automatically with full satisfaction. In 2025, generative AI is turning this point of friction into a competitive advantage.
The Observation: A Blind Spot in Policyholder Relations
The Benefits Table: A Sticking Point in Insurance
“What exactly am I covered for?” This is a question all insurers hear on a daily basis. And for good reason: according to a 2024 UFC-Que Choisir study, 78% of policyholders admit they have not read their policy in full, and 82% acknowledge they do not understand the terms of their benefits table.
This observation is not merely a communication problem. It has tangible business impacts:
- Increase in claims: policyholders discover exclusions at the time of a claim
- Customer service overload: up to 40% of tickets relate to understanding coverage
- Erosion of trust: a Net Promoter Score (NPS) on average 15 points lower than in other service sectors
- High operational cost: handling a simple request keeps an advisor occupied for 8 to 12 minutes on average
The paradox is striking: insurers invest heavily in designing sophisticated products, yet fail to make them understandable at the moment when the policyholder needs them most.
The Limitations of Traditional Solutions
Faced with this challenge, the industry has tried many approaches:
- Static FAQs quickly become outdated. They never cover all edge cases and force the policyholder to navigate a complex decision tree. The result: 73% of users give up before finding their answer.
- Rule-based chatbots deliver a frustrating experience. Unable to understand the nuances of natural language, they systematically redirect users to an advisor the moment a question falls outside predefined scripts. The abandonment rate often reaches 60% after two unsuccessful interactions.
- Advisors, overwhelmed, spend 65% of their time on level-1 questions that require no specific analysis. This situation creates a double loss: high cost for the insurer and frustration among teams who cannot focus on high-value advisory work.
Faced with this situation, a new generation of solutions is emerging, driven by the spectacular advances of generative artificial intelligence.
Generative AI: A Disruptive Technology for Policyholder Understanding
How Does Conversational AI Work Applied to Insurance?
Generative artificial intelligence represents a fundamental break from traditional chatbots. Where the latter operate on rigid decision trees (if question X, then answer Y), generative AI genuinely understands the meaning of the customer’s request.
The approach rests on three technological pillars:
First, advanced natural language understanding makes it possible to interpret questions even when formulated imprecisely or using non-technical terms. “My child is going to summer camp this year, are they covered?” will be understood as a query about the extent of assistance coverage for a minor travelling on a school trip. The AI agent is also capable of asking follow-up questions to clarify the initial request and thereby refine the final response.
Second, document semantic analysis goes beyond simple keyword search. The AI ingests the entire contractual documentation (general conditions, specific conditions, benefits tables, endorsements) and extracts meaning from it. It establishes connections between different clauses and can respond to questions requiring the synthesis of multiple sources.
Finally, contextual generation produces personalised responses. The insurer transmits policyholder information to the AI while respecting data confidentiality and sovereignty. The conversational agent then processes the request based on the policyholder’s profile, their specific policy, and even the conversation history, without needing to ask for specific policy details. Each response cites the relevant contractual sources precisely, ensuring transparency and compliance.
This approach differs radically from traditional solutions: rather than trying to anticipate every possible question, the AI adapts to each unique request, in the customer’s own language.
The 3 Key Capabilities of an AI Agent for Insurance
1. Intelligent Reading of Contractual Documents
A modern AI agent no longer simply searches for keywords in a PDF. It analyses the logical structure of the contractual document (PDF, Excel, or other format), understands the relationships between main guarantees, extensions, and exclusions, and identifies the applicable conditions.
Consider a concrete example: a benefits table for optical coverage mentions “€150/year with renewal possible every 2 years for prescription lenses”. The AI understands that:
- The annual ceiling is €150
- Early renewal is possible but subject to a time condition
- This coverage applies only to lenses, not to frames (unless stated otherwise elsewhere in the policy)
This analytical capability makes it possible to respond to questions such as “I broke my glasses 18 months ago, can I be reimbursed?” with a precise and sourced answer.
2. Personalised Responses in Real Time
The strength of generative AI lies in its ability to tailor each response to the specific context.
First scenario: “I’m going to Thailand for 3 weeks, am I covered?” The AI analyses the customer’s home insurance policy, identifies the holiday coverage, verifies the maximum duration (often 90 days), and confirms coverage while specifying the ceiling and the steps to follow in the event of a claim.
Second scenario: “My 22-year-old son still lives at home, is he on my policy?” The AI checks the maximum age for inclusion (generally 25 if studying, otherwise 21), the tax status, and provides a personalised response with a precise reference to the relevant article of the policy.
This personalisation extends to the format: the AI adapts its language level, can provide concrete numerical examples, and systematically suggests the next steps (documents to provide, form to complete, emergency contact number).
Round-the-clock availability transforms the experience: no more waiting, no more limited time slots. The policyholder receives their answer immediately, whether at 2pm or 11pm on a Sunday.
3. Continuous Learning and Improvement
Unlike static systems, an AI agent evolves constantly. Each interaction feeds an improvement loop:
- Detection of areas of misunderstanding: if 15% of policyholders ask questions about the excess in the event of glass breakage, this signals unclear contractual wording or a communication gap.
- Knowledge base enrichment: new legitimate questions that do not receive an immediate response are escalated to the business teams, then integrated after validation.
- Continuous optimisation: conversation analysis makes it possible to identify effective phrasing, examples that resonate with policyholders, and progressively refine the relevance of responses.
This learning dimension makes the AI not merely a response tool, but a genuine observatory of the customer experience that generates strategic insights for improving the products themselves.
Concrete Use Case: Genii and the Simplification of Coverage
The Challenge Addressed by tolk.ai
Consider a general French insurer offering different policies (car, home, health, life), each with multiple coverage levels and modular options. The result: hundreds of possible combinations of coverage.
The initial situation:
- 1,200 tickets per week to customer service
- 42% relate to understanding coverage (level-1 questions)
- Average handling time: 9 minutes per ticket
- Estimated cost: €180,000 annually to handle these simple requests
- Customer satisfaction on this channel: between 60 and 70%
The objective: to intelligently automate these level-1 requests while improving the customer experience.
The result with Genii: 85% of requests resolved automatically with 95% user satisfaction, representing a radical transformation of the customer service.
How Genii Works in a Real-World Setting
Scenario 1: The Pre-Subscription Customer
Customer question: “My 19-year-old daughter is a student but still lives with me for 6 months of the year. If she has an accident during her internship abroad, is she covered by my home insurance or does she need to take out her own policy?”
Processing by Genii:
- Profile analysis: detection of the situation “child fiscally attached, higher education”
- Multi-document consultation: verification in the general conditions + assistance benefits table
- Structured response:
“Yes, your daughter is covered by your Home Multirisque policy under the assistance coverage. Here are the details:
Eligibility: Children up to age 25 who are pursuing their studies are covered even if they reside elsewhere for part of the year (Article 3.2 of your General Conditions)
Coverage abroad: Assistance applies for stays of up to 90 consecutive days, including professional internships (Benefits Table, Assistance section)
Benefits included: medical expenses up to €50,000, medical repatriation, advance of hospitalisation costs
⚠️ Point of attention: For an internship lasting more than 3 months, a coverage extension is recommended. Would you like a quote for this option? In case of emergency: +33 1 XX XX XX XX (available 24/7)”
Impact: The customer receives a complete, reassuring, and immediately actionable response. Genii also identifies a relevant upsell opportunity, without being intrusive.
Scenario 2: The Customer Facing a Claim
Customer question: “I just broke my glasses while playing sport, how much am I entitled to?”
Processing by Genii:
- Policy verification: identification of the optical coverage subscribed (level 2)
- History consultation: last optical reimbursement 14 months ago
- Personalised calculation:
“Based on your Santé Confort+ policy, here are your reimbursement entitlements:
📋 Your optical coverage: – Annual ceiling: €200 for lenses + frame – Renewal possible every 2 years (except in cases of vision change or accidental breakage)
💶 In your situation: – Last optical reimbursement: February 2024 (€142) – You are eligible due to accidental breakage – Maximum amount available this year: €200
📎 For your claim file: 1. Detailed optician invoice 2. Optometrist medical certificate (prescription) 3. Upload these documents via your customer account or send them to remboursements@votreassureur.fr
⏱️ Standard processing time: 7 business days. Need help putting together your claim file?”
Impact: The customer has immediate visibility over their entitlements, the amount, and the steps to follow. No surprises, no frustration from misunderstanding their coverage.
Scenario 3: The Complex Question Requiring Escalation
Customer question: “I have water damage at my property, my upstairs neighbour says it’s not their responsibility but I think it is. Who should file the claim?”
Processing by Genii:
- Complexity detection: legal question requiring liability analysis
- Level-1 response:
“Your question involves determining liability, which requires an analysis of your specific situation. Here is what I can tell you:
✓ Your policy protects you: Your Water Damage coverage covers damage at your property, regardless of who is responsible
✓ Immediate next step: You must file a claim within 5 days. An assessor will be appointed to establish liability
📞 I am connecting you with an advisor who will analyse your situation and guide you on the legal aspects. Are you available now or would you prefer a callback?”
- Intelligent transfer: creation of a ticket with full context for the advisor
Impact: Genii recognises its limitations, still provides useful information, and ensures a smooth handover to a human without the customer having to repeat themselves.
Measurable Results
After several months of progressive deployment, the metrics speak for themselves:
📊 Operational Efficiency
- 85% automatic resolution: out of 1,200 weekly requests, 1,020 handled by Genii
- Average resolution time: 45 seconds (vs 9 minutes with an advisor)
- Estimated savings: €153,000 per year on level-1 handling
- Team refocus: the remaining 180 requests benefit from 15 minutes of handling time (vs 9 minutes previously) for premium advisory support
📈 Satisfaction and Customer Experience
- 95% user satisfaction on interactions with Genii (vs 71% by phone)
- NPS increase of +12 points among customers who use Genii
- Adoption rate: 68% of customers use Genii as their first choice for questions (after 3 months)
- Waiting time: 0 seconds vs an average of 4 minutes 30 seconds by phone
🔍 Strategic Insights
- Identification of 14 recurring areas of misunderstanding which led to the redesign of 3 benefits tables
- Detection of 23 legitimate questions not covered by the documentation, now integrated
- Analysis of peak periods enabling better allocation of human resources
The Generative AI Revolution: Tangible Benefits
Implementing a conversational AI solution like Genii delivers measurable benefits for the entire ecosystem.
For the Policyholder
Immediate autonomy without frustration No more endless waiting times, no more transfers from one department to another, no more ‘your call is important to us’ on repeat. The customer receives a response in under a minute, 24/7, including public holidays.
Clear understanding of their coverage Responses are no longer impenetrable legal jargon but personalised explanations, with concrete examples and specific amounts. The customer knows exactly what they are entitled to, without ambiguity.
Strengthened trust in their insurer Complete transparency (every response cites the relevant contractual clauses precisely) and the accuracy of information create a sense of control. The customer no longer feels they are signing up ‘blindly’.
Universal accessibility Whether the customer is comfortable with digital tools or prefers a simple interface, whether they have a question at 3pm or midnight, the AI adapts and remains accessible. This also includes people with disabilities (screen reader compatibility, language level adjustment).
For the Insurer
📊 Drastic reduction in operational costs With 85% automatic resolution, the savings are immediate: a phone call costs between €6 and €12 according to sector studies. Across 50,000 automated interactions per year, the saving amounts to between €275,000 and €575,000.
📈 Spectacular NPS improvement Achieving 95% satisfaction on interactions is not merely an image issue. Each NPS point gained statistically translates into a 0.5 to 1% reduction in the churn rate. Across a portfolio of 100,000 customers with an average premium of €800, this represents between €400,000 and €800,000 in retained premiums.
🎯 Intelligent up-sell and cross-sell Unlike generic marketing campaigns, Genii detects real needs in context. A policyholder asking whether their electric bicycle is covered is offered the relevant extension at precisely the right moment. Conversion rate: 4 to 7 times higher than standard campaigns.
🔍 A goldmine of data and insights Each interaction generates structured data: types of questions, moments of confusion, documentation gaps, unmet expectations. These insights enable continuous improvement of products, communication, and anticipation of regulatory developments.
🚀 Competitive differentiation In an increasingly commoditised market where products look alike, the customer experience becomes the primary differentiating factor. Offering a genuinely innovative service attracts new policyholders and retains existing ones.
⚡ Infinite scalability An advisor handles 4 to 6 calls per hour. Genii can manage hundreds of simultaneous conversations without any degradation in quality. The ability to absorb peaks (following a weather event, at a new product launch) without temporary recruitment is a major asset.
For the Teams
Revaluation of the advisor role Freed from repetitive level-1 questions, teams can focus on what makes human expertise valuable: empathy in difficult moments, analysis of complex situations, personalised advisory support for premium policyholders.
Reduction of stress and improvement in quality of working life Answering ‘yes, your excess is €200’ fifty times a day is demotivating. Supporting a policyholder after a traumatic claim with the time needed is rewarding. Internal surveys show a 31% increase in advisor satisfaction following the deployment of Genii.
Natural skills development Advisors now handle predominantly cases that require expertise and judgement. This qualitative evolution develops skills and makes roles more attractive to young talent.
Human-machine collaboration Genii is not a competitor but an assistant. Advisors benefit from the same document analysis tools to handle complex cases more efficiently. Time saved on information retrieval is reinvested in the relationship.
Conversational AI: A Strategic Investment for 2026
We are at a tipping point in insurance customer relations. Generative AI is no longer a laboratory experiment but an operational reality, successfully deployed by pioneering players who are already reaping the rewards.
The figures speak for themselves: 85% automatic resolution with 95% customer satisfaction. These results are now reproducible at scale.
Insurers who adopt conversational AI now are gaining a decisive head start across three strategic dimensions:
First, the policyholder experience, which is becoming the battlefield for differentiation. In a market where products are increasingly similar, the ability to respond instantly, clearly, and relevantly to customer questions becomes a major competitive advantage. Policyholders who have experienced Genii do not want to go back.
Then, operational efficiency, in a context of margin pressure. Automating 85% of level-1 requests frees up considerable resources, which can be reallocated to premium advisory support, business development, or simply to improving profitability. The business case is clearly positive from the first year.
Also, in a sector that struggles to attract young talent, AI enhances employer attractiveness. Offering employees access to cutting-edge AI tools and the ability to focus on high-value tasks rather than repetitive ones is an increasingly decisive recruitment and retention argument.
Furthermore, the benefits table, long perceived as an unreadable regulatory constraint, becomes, thanks to AI, an asset for transparency and trust. This technical document, often a source of anxiety for the policyholder, is transformed into a natural and reassuring conversation. ‘Am I covered for X?’ finally receives a clear, immediate, and personalised answer.
This transformation is not reserved for digital pure players or sector giants. Solutions like Genii are designed to adapt to organisations of all sizes, with progressive deployment models that limit risk.
In conclusion, the question is no longer ‘should we’ but ‘how and when’ to integrate conversational AI into your policyholder relations strategy.
