Concrete, measurable, and adopted AI solutions — from framing to production.
Sirene
Natural language company search.
Business Challenge
Sales and operations teams waste valuable time identifying and qualifying companies through the standard INSEE Sirene interface. Rigid filter syntax and lack of semantic understanding hinder adoption.
AI Solution
An AI assistant that understands natural language queries and interrogates the INSEE Sirene API to return structured, filtered, and enriched results — with no technical knowledge required from the user.
Key Features
Natural Language Queries
Automatic translation of NL queries into Sirene API parameters via a finely calibrated language model.
Semantic Filters
Intelligent segmentation by NAF sector, region, company size and legal status from free-form expressions.
Enriched Export
Dashboard generation and CSV/JSON exports with enriched and normalized SIRET data.
Tech Stack
Comex
Strategic framing and alignment through AI.
Business Challenge
Executive committees face a growing volume of AI initiatives without structured tooling to prioritize, frame, and align them with company strategic objectives in a controlled timeframe.
AI Solution
An AI-powered strategic deliverable generator, structuring Comex discussions around proven management frameworks (OKR, McKinsey 7S, Porter) to accelerate decision-making.
Key Features
AI Roadmap Generation
Automatic creation of 12-18 month AI roadmaps, prioritized by business value and effort level.
Automatic OKR Alignment
Mapping of AI initiatives onto existing company OKRs with alignment scoring and conflict detection.
Sector Benchmarks
Instant positioning against industry AI practices via a database of analyzed case studies.
Tech Stack
DVF
Real estate transaction search and analysis.
Business Challenge
Analyzing DVF (Property Value Requests) data published by the DGFiP is tedious: multi-GB raw CSV format, no native visualization, and difficulty cross-referencing with geographic data.
AI Solution
A real estate transaction search and analysis platform integrating interactive mapping, statistical zone analysis and AI insight generation from DGFiP open data.
Key Features
Transaction Mapping
Interactive geographic visualization of all real estate transfers, with intelligent density clustering.
Price/m² Analysis by Zone
Automatic calculation of price medians by district, municipality and department with time-series trends.
AI Trend Detection
Identification of high price-movement zones and market anomalies via predictive analytics models.
Tech Stack
Fraud Agent
Agentic fraud detection and recovery pipeline for property & casualty insurance.
Business Challenge
Fraud teams at P&C insurers are overwhelmed by false positives from fixed rule-based alerts, while subrogation recovery claims against liable third parties are identified too late, or missed entirely, due to lack of time to cross-reference cases.
AI Solution
A 5-stage agentic pipeline that scores each incoming claim, detects fraud networks through graph analysis, reads police reports using generative AI, and automatically triggers relevant recovery cases — a demo built on a fictional dataset, illustrating a multi-service agentic architecture on Azure.
Key Features
Explainable Scoring
0-100 fraud score per claim with Reason Codes, combining a Machine Learning model with business rules.
Network Detection
Graph-based analysis of cross-case links (repair shops, IP addresses, grouped filings) invisible in single-case review.
Automated Recovery
Police report reading via generative AI and automatic generation of the subrogation case file.
Tech Stack
ID Document Verifier
AI-powered document quality control and MRZ consistency check.
Business Challenge
A subscription file with an illegible or incomplete ID gets stuck until a handler notices it and contacts the client — a back-and-forth cycle that degrades completion rates and burdens the back office.
AI Solution
A demonstration that checks the quality of an identity document at upload time (legibility, completeness, framing) and verifies the mathematical consistency of its MRZ zone against ICAO 9303 — without ever judging its authenticity.
Key Features
Instant quality control
Detection in seconds of blurry photos, glare, or partially out-of-frame documents, before a file ever reaches a handler.
Verifiable MRZ consistency
Deterministic calculation of ICAO 9303 check digits for ID cards and passports, with the calculation detail displayed — not a black box.
Front/back cross-check
Comparison of the document number between both sides of the card, as an additional consistency signal.
Tech Stack
LexGuard
Detects red-flag signals in contracts before signature.
Business Challenge
The mayor of a rural town, with no legal department, signs a telecom equipment contract under sales pressure — one that conceals a second leasing agreement with a long commitment term and near-impossible termination. Some towns have been sued for tens of thousands of euros after the fact.
AI Solution
A demonstration that analyzes a contract submitted as a PDF and flags, before signature, the warning signs of a risky arrangement (concealed contract, abnormal duration, tacit renewal) — in plain language, without ever ruling on the contract's legality or a company's reliability.
Key Features
Red-flag signal detection
Identifies patterns of risky contractual arrangements.
Plain-language verdict
3-level assessment, no legal jargon.
Zero data retained
Ephemeral processing, no document or result is stored.