Transforming Insurance Communication with DIALOGIC
AI-Driven Clarity, Personalisation, and Compliance for Accessible, Trustworthy Contracts Across Europe’s Digital Insurance Ecosystem
DIALOGIC is a European research and innovation project advancing how complex documents are written, explained, and understood.
The project combines two strands of cognitive science with machine learning: Paul Grice’s theory of cooperative communication, and recent work on Theory of Mind in large language models. The goal is a working platform that gives writers real-time, explainable feedback on the clarity of what they produce, and gives readers personalised explanations of what they receive.
DIALOGIC builds directly on results from the EIC Pathfinder project ASTOUND, advancing the technology from laboratory proof-of-concept (TRL 3–4) to a system validated in real-world insurance environments (TRL 6) over 24 months, from April 2026 to March 2028.
The project is carried out by Indeep Artificial Intelligence as sole beneficiary.
OBJECTIVES
DIALOGIC’s two-year work plan is organised around two intertwined ambitions: building the technology, and proving it in the field.
Technological development
On the technical side, the project advances the linguistic-pragmatic engine and the SaaS editor inherited from ASTOUND to production readiness, so that writers receive real-time, explainable feedback on the clarity of what they produce, grounded in measurable linguistic criteria. Alongside the editor, DIALOGIC integrates a Theory-of-Mind module with a browser-based assistant capable of adapting explanations of document content to the reader’s likely needs, prior knowledge, and context. Underpinning both is a commitment to operating within the European regulatory perimeter: the project’s security, privacy, and accessibility work targets compliance with ISO 27001, EN 301 549, the GDPR, the AI Act, and the Insurance Distribution Directive, not as a post-hoc audit exercise but as a design constraint from the outset.
Validation and deployment
The second strand turns the technology outward. DIALOGIC will run structured pilots in the insurance sector measuring comprehension, regulatory clarity, and operational efficiency against pre-defined baselines, so that the platform’s value is demonstrated quantitatively rather than asserted. From this validated base, the project will reach commercial readiness in at least two European markets, with the platform localised, scaled, and engaged with early enterprise adopters. In parallel, DIALOGIC establishes a defensible intellectual property position, exploring potential patentability while publishing an open specification for the project’s document-quality label, on the principle that public standards and protected inventions are complementary rather than opposed.
SCIENTIFIC THEORY
DIALOGIC rests on two complementary bodies of work in the cognitive sciences.
Grice’s cooperative principle
In a series of lectures delivered at Harvard in 1967 and published as Logic and Conversation, the British philosopher Paul Grice proposed that everyday communication is governed by an implicit cooperative principle: speakers and writers tacitly commit to making their contributions truthful, sufficiently informative, relevant to the exchange, and clearly expressed. Grice formulated these commitments as four maxims — Quantity, Quality, Relation, and Manner — that together describe what it means for a message to be cooperative in context.
Most institutional documents — insurance policies, terms of service, regulatory notices — fail one or more of these maxims by construction. They are written for legal robustness, not for the reader. DIALOGIC’s linguistic-pragmatic engine operationalises Grice’s maxims as measurable criteria, and surfaces violations as actionable feedback during drafting.
Theory of Mind in language models
Theory of Mind (ToM) refers to the cognitive capacity to attribute mental states, such as beliefs, intentions, and knowledge to others, and to reason about how those states differ from one’s own. It is the foundation of any explanation tailored to a particular listener.
In a study published in Nature Human Behaviour in 2024, Strachan, Becchio, and colleagues demonstrated that contemporary large language models perform at or above human levels on a battery of standard Theory of Mind tasks, while exhibiting characteristic patterns of error that distinguish them from human reasoners (Strachan et al., 2024).
DIALOGIC builds on this result to drive a reader-side assistant that infers what a given reader is likely to find unclear, ambiguous, or missing, and adapts its explanations accordingly.
IMPACT
The case for DIALOGIC rests on four convergent arguments.
Economic
Unclear institutional documents generate measurable costs: customer support queries, complaints, legal review cycles, and missed regulatory deadlines. The European contract-intelligence market within insurance alone is projected to reach approximately €35 billion by 2030. DIALOGIC targets this market with a differentiated proposition: clarity at the source, rather than remediation downstream.
Regulatory
Recent and forthcoming European instruments like the GDPR, the AI Act, the Insurance Distribution Directive (IDD), the European Accessibility Act, and more, increasingly require that information addressed to consumers be intelligible, accessible, and explainable. DIALOGIC operationalises these requirements as engineering criteria.
Societal
Information asymmetry between institutions and the people they serve is a primary driver of digital exclusion. Vulnerable populations — older adults, migrants, people with cognitive or sensory disabilities — pay the highest price for documents written without them in mind. DIALOGIC aligns with UN Sustainable Development Goals 4 (quality education), 10 (reduced inequalities), and 16 (peace, justice, and strong institutions).
Scientific
DIALOGIC contributes to ongoing research on pragmatic competence in large language models, Theory of Mind in artificial agents, and the measurement of document quality at scale. The project commits to open-access publication of its scientific outputs under Creative Commons licensing where compatible with the protection of patentable subject matter.
CONTACT US
For project information, partnership enquiries, or media requests, contact the DIALOGIC project team. We respond within one working day.