The NCE — Neural Context Engine — is not an AI laboratory project. It is a precise answer to a precise pain point — built by four complementary profiles who have experienced, from the inside, the same problems you face.
Cybersecurity, data science, development and commercial — KnowWeave was born from the meeting of complementary skills around a shared observation: regulated industry deserves better than generic RAG.
Guardian of platform sovereignty and containment. His cybersecurity experience at La Poste (consumer messaging) trained him in on-premise architecture and rigorous containment mechanisms — he designs the barriers that ensure your data never leaves your infrastructure.
Architect of the ingestion pipelines and the NCE AI core. Her background in Transport & Logistics gives her a concrete understanding of industrial document complexity. She translates your document corpora into structured knowledge networks — with the rigour of a data scientist and the precision of an engineer.
He understands your problems before talking about the product. His experience in the phytosanitary sector gives him the domain credibility to speak with CIOs, regulatory departments and executive directors — in their vocabulary, not technology jargon. Your first point of contact at KnowWeave.
More than 25 years leading IT projects in document-intensive sectors: phytosanitary, agrochemicals, automotive, aerospace (Airbus), motorcycle, purchasing centre, nuclear energy (La CLI). He lived the pain of unmanageable archives from the inside. The NCE is the answer he wished he had.
"The knowledge was there — it just wasn't accessible."
Five months from the initial conviction to launch. Each step grounded in an observation, a decision, a deliverable.
Olivier ALESSANDRI formalises the idea of a neural context engine for regulated industry. Core finding: generic RAG solutions are not fit for environments subject to audit, traceability and strict regulatory compliance.
Design of the Neural Context Engine architecture: on-premise Knowledge Graph, regulatory entity extraction by domain, zero-data-leak inference engine. The four co-founders unite around the project.
Prototype developed and tested on a real regulatory dataset. Approach validated: the NCE semantic knowledge network outperforms classic vector RAG on complex multi-document queries.
Definition of the four-step deployment methodology. Construction of the offer, contractual framework (mutual NDA, measurable deliverables, result guarantees) and pricing grid.
Official creation of KnowWeave. Four co-founders. The product is the NCE — Neural Context Engine. The website is under construction.
Website live, offer available. The tool keeps evolving, driven by improvements delivered continuously. The promise remains the same: 12 weeks, measurable precision, total sovereignty.
No announcements for effect. Three structural axes for upcoming releases, communicated as they are delivered.
NCE spots and flags inconsistencies in your documents — values out of range, outdated items, contradictory sources — and only admits into its memory information that is verified and tied to a recognised source.
in developmentOpening up to other regulated sectors beyond phytosanitary, answers adapted to each user's role, and advanced multilingual command. On the horizon: autonomous investigations run across your corpus.
in developmentCompare network state between two dates, track the evolution of a standard, and long conversations that maintain network context with proactive suggestions of inconsistencies.
plannedA faster but non-traceable answer does not interest us. Every output must be defensible.
100% on-premise. Not as an option, not as an add-on. It is the baseline, not the luxury.
If your case is not a fit, we tell you. We prefer to lose a contract than to disappoint you.
Our figures are measured and auditable. No marketing slide with invented percentages.
Not a one-shot supplier. The contract says it explicitly: we accompany you over time.
Mutual NDA proposed from the first contact. Your data serves nothing other than you.
30 minutes. Your brief. Our reading. The next step, you decide.