Semantic Forge
Give your AI a map of your business.
Our Smart Semantic Layer reads your databases and automatically builds a formal ontology mapping the databases. Through MCP, any compatible AI assistant can reason over that model to find exactly the right data, while the data itself never leaves your control.
Allow the assistant to understand:
- the in-house vocabulary used
- the relationship between business concepts
- how to retrieve information precisely and reliably
While retaining control over which information is shared and with whom.
LLMs are fluent. Business data isn't.
Organisations want AI that answers questions about their own business. But most business knowledge lives in structured databases that an LLM can't reliably read, and shouldn't be allowed to see in full.
Meaning is missing
Table and column names like CUST_TXN_HDR say little about what the data means. That knowledge sits with a few key people.
Text-to-SQL guesses
Asked to write queries over unfamiliar schemas, models invent joins and misread relationships. The answers look plausible and are wrong.
Data Security
Pasting records into prompts or copying them into vector stores creates governance, privacy and compliance risk.
The AI reasons over meaning. The Smart Semantic Layer handles the data.
The assistant never queries your databases directly. It explores a formal model of your business through MCP tools and composes a description logic query in readable OWL Manchester syntax. The Smart Semantic Layer runs that query in place, under the user's own permissions.
Build the ontology automatically
The OWL ontology is generated directly from the source databases: their models, data dictionaries and comments, across multiple database types. Optionally, an administrator can further categorise content and map it to the organisation's own business vocabulary.
Infer how things connect
OWL RL reasoning with data extensions works out where customers, products, sales and any other relevant business data live and how they relate, across every connected source.
Let the AI query by meaning
Through MCP, the assistant explores the model and builds a precise query. The Smart Semantic Layer runs it securely and returns only what the AI needs.
Grounded in logic, not retrieval tricks.
Automated, not hand-built
Ontology engineering usually means months of specialist effort. Here the model is read directly from your database sources, so the work left is categorising it, not building it.
Model-agnostic
Built on the open Model Context Protocol, so it works with any compatible assistant or agent framework rather than a single vendor.
Formal semantics
An OWL ontology with RL reasoning gives the AI a logically sound model to query, not text chunks or inferred guesses about a schema.
Private by design
Data stays where it is, queried in place, with no copies in lakes or vector stores. The AI only sees structure and result counts; records are shown only to authorised users in the Smart Semantic Layer.
Permission-aware
Every query runs under the existing security model, with group permissions, user logging and an auditable history.
Explainable
Each answer traces back to named business concepts and an inspectable description logic query in Manchester syntax, readable by people as well as machines, so results can be checked and trusted.
Bring semantic grounding to your AI offering.
We're looking for a small number of partners to take the Smart Semantic Layer further.
Technology partners
Data platforms, AI and analytics vendors can embed the Smart Semantic Layer as the semantic grounding behind their assistants and agents, or incorporate it in their product suites.
Integrators & consultancies
Deliver governed, accurate AI on client data faster, with a semantic model that keeps sensitive records out of the prompt.
Licensing & strategic investment
We welcome conversations about licensing, OEM arrangements and strategic investment in the platform and its intellectual property.
See the demo.
Watch an AI assistant answer real business questions through the Smart Semantic Layer. Tell us your organisation and role, and we'll arrange a session.