Read Every Page. Flag Every Risk.
Construction Tenders in France.

The State of Construction Procurement in France

Updated 15 May 2026

## Extracting the CCAG-Travaux Compliance Matrix via Gemini When parsing a 400-page Dossier de Consultation des Entreprises (DCE) published on the BOAMP, manual extraction of mandatory deliverables often misses buried environmental certifications. Lucius AI utilizes a Gemini-extracted compliance matrix to instantly map every requirement stipulated under the Code de la commande publique Article L2112-2. For example, during a €14.5M hospital renovation tender issued by the Assistance Publique – Hôpitaux de Paris (AP-HP) in October 2023, the system isolated 47 distinct technical prerequisites from the Cahier des Clauses Techniques Particulières (CCTP). When dealing with multi-lot contracts under the Allotissement rules of the Code de la commande publique, the AI separates the compliance matrix by specific trade, ensuring the Gros Œuvre (Lot 1) requirements do not bleed into the Menuiserie Extérieure (Lot 4) deliverables. The Files API caching mechanism stores these parsed CCTP requirements, ensuring the tender writer's matrix remains synchronized even if the buyer issues an updated Avis Rectificatif 48 hours before the deadline. This automated extraction directly links each required NF EN 1992-1-1 Eurocode 2 concrete specification to the exact page number in the source PDF, preventing non-compliant material proposals. By anchoring the matrix to the specific CCAG-Travaux 2021 framework version, the platform guarantees that all mandatory site preparation deliverables are accounted for before drafting begins.

## Identifying Pénalités de Retard and Asymmetric Indemnities in the CCAP Construction contracts governed by the Code de la commande publique frequently embed severe financial risks within the Cahier des Clauses Administratives Particulières (CCAP). Lucius AI deploys targeted risk flag detection to highlight deviations from standard CCAG-Travaux liability caps, specifically scanning for asymmetric indemnities favoring the public buyer. In a recent €8.2M municipal tramway extension project for Bordeaux Métropole, the system flagged a hidden clause imposing €5,000 daily pénalités de retard that lacked the standard 10 percent contract value ceiling. If the buyer attempts to shift the burden of geological risk onto the contractor—a direct violation of the standard Norme NF P 03-001—the system isolates the offending paragraph for immediate review. The platform's Deep Think contradiction audit cross-references these penalty clauses against the Acte d'Engagement (ATTRI1 form) to ensure the financial exposure matches the officially signed commitment. Tender writers receive an immediate alert if the buyer's CCAP attempts to override the mandatory insurance thresholds dictated by the Loi Spinetta for decennial liability. This allows the drafting team to formulate precise technical qualifications or request formal clarifications via the buyer's portal before the standard 6-day Q&A cutoff.

## Deep Think Contradiction Audits Across the DCE Public construction tenders in France notoriously suffer from misaligned specifications between the architectural plans and the written CCTP. Lucius AI executes a Deep Think contradiction audit across the entire DCE pack, analyzing the Règlement de la Consultation (RC) against the Bordereau des Prix Unitaires (BPU). Consider a €22M secondary school construction RFP published by the Région Île-de-France in January 2024; the architectural drawings mandated RE2020-compliant biosourced insulation, while the BPU only provided pricing lines for standard polyurethane foam. Furthermore, the system cross-checks the Décomposition du Prix Global et Forfaitaire (DPGF) against the CCTP to ensure no mandatory structural steel elements are omitted from the final pricing. The Files API caching system holds the entire 2GB drawing set and the Excel pricing sheets in active memory, allowing the AI to pinpoint this exact material discrepancy instantly. By identifying these clause-vs-clause contradictions early, the tender writer can submit a targeted question to the Direction de l'Immobilier de l'État (DIE) rather than submitting a non-compliant pricing schedule. The audit also verifies that the execution deadlines stated in the RC perfectly match the phasing requirements detailed in the Plan Général de Coordination (PGC).

## Drafting the Mémoire Technique Using File Search Citations Constructing a high-scoring Mémoire Technique requires precise alignment with the buyer's grading criteria, which are strictly regulated under Article R2152-7 of the Code de la commande publique. Lucius AI powers draft generation grounded in the bidder's past won responses, utilizing File Search citations to pull exact methodologies from previously successful bids. For a €5.4M social housing retrofit issued by Paris Habitat, the system automatically extracted the exact asbestos removal protocol (Sous-section 4) that won a similar contract in Lyon the previous year. The system also integrates specific site manager CVs formatted to the exact requirements of the buyer's Cadre de Réponse, ensuring the qualifications match the OPQIBI certifications demanded in the RFP. The AI weaves these proven technical narratives directly into the new draft, appending File Search citations that link back to the specific SOPs stored in the company's SharePoint repository. Furthermore, the Gemini-extracted compliance matrix ensures that the generated text explicitly addresses the mandatory carbon footprint reduction targets specified in the buyer's Charte Chantiers à Faibles Nuisances. This guarantees the drafted response relies on verified, company-specific engineering data rather than generic construction methodologies, directly satisfying the stringent evaluation grids used by the Commission d'Appel d'Offres (CAO).

## Validating DUME and PLACE Plateforme des Achats Submission Readiness The final hurdle in French public procurement is navigating the rigid administrative requirements of the Document Unique de Marché Européen (DUME) and the electronic submission portals. Lucius AI performs a comprehensive submission readiness check against the buyer's stated rules, ensuring every required form, from the DC1 to the DC4 subcontracting declarations, is present and correctly signed. During a €31M infrastructure upgrade for SNCF Réseau, the platform's final audit detected that the required Kbis extract was 95 days old, violating the strict 3-month validity rule mandated by the RC. Before the final ZIP file is generated, the platform verifies that the mandatory Attestation de Régularité Fiscale is present and dated within the required timeframe. The system specifically verifies that the digital signature applied to the Acte d'Engagement complies with the RGS** (Référentiel Général de Sécurité) standard required by the PLACE plateforme des achats. By cross-referencing the final compiled PDF package against the Gemini-extracted compliance matrix, the AI confirms that the file size does not exceed the 50MB per-document limit imposed by the local procurement portal. This rigorous validation prevents technical disqualifications at the point of upload, ensuring the meticulously drafted Mémoire Technique actually reaches the evaluation committee of the Ministère de la Transition Écologique.

Bidders into France construction contracts compete under BOAMP, PLACE and the French Code de la commande publique. Sector-specific compliance bars include CDM 2015, JCT/NEC4 form selection, retention bonds, social value and net-zero commitments — Lucius AI maps each one to your response with a page-cited audit trail, so legal review reads as fast as engineering review.