How it works
From scattered files to answers anyone in the organization can trust, in three steps.
- Connect the sourcesDocuments, spreadsheets, databases and shared folders are gathered in one place and kept up to date automatically. No manual copying, no duplicate archives.
- Ask in plain languageAsk the way you would ask an expert colleague. Every answer comes with the documents it is based on, one click away.
- See and share the knowledgeExplore information as a map of connections, a table or a view built for a specific team, and share it with colleagues or embed it on a website.


More than traditional RAG
Traditional retrieval-augmented generation selects content by semantic similarity to the question. Rational AI combines semantic search with a knowledge graph: it retrieves the most relevant resources, extends the search along the relations between them, and ranks the results by relevance. Every step is recorded and visible in the answer.
- Initial retrievalThe question is compared with the content of the knowledge base by meaning, not by keywords, so “late deliveries” also finds “shipping delays.” Resources you select yourself in the chat are always included, even when similarity alone would leave them out.
- Graph expansionStarting from the strongest matches, the search follows the relations in the knowledge graph (a product to its datasheet, a datasheet to the orders that use it) and, when enabled, the tags resources share. This brings in information that answers the question without resembling it.
- SelectionEach resource gets a relevance score that weighs meaning against exact terms. Resources below the threshold are discarded, and the conversation shows how many were kept, how many were discarded, and why.



The right interface for each process
The same knowledge can be read as a graph, a table, a map, a timeline, a card wall or a chart. A view is generated from a description, using the knowledge's real attributes and categories. Technical teams can also build views in HTML, React or as Web Components.

Augment documents with annotations and relationships
Each piece of information is augmented with its context: a number becomes a date, an image has its content recognized, a name is associated with a role.

Every claim checked against the sources
Every answer shows the source behind each claim, so verifying any piece of information is always one click away.

Knowledge always up to date
Sync with your data sources is automated and your knowledge stays up to date with every change.
Your knowledge base, on any website
Give customers, partners or employees access to your knowledge through custom touchpoints: support chats, help centers, product configurators, interactive views and more.
Rational Knowledge transforms fragmented data into a dynamic, explorable knowledge network.
It consolidates information from documents, databases, and applications into a Knowledge Graph — empowering teams to search, visualize, and interact with data intuitively and transparently.

Knowledge Graph
Go beyond traditional retrieval with Graph RAG, turning search into knowledge mapping.
Visualize and navigate all your organizational knowledge in an interactive, graph-based view.
Discover semantic relationships among documents, people, and concepts, and interact via chat to retrieve, connect, or update knowledge in real time.
Generate answers that are:
- Relevant, by following relationships instead of keywords
- Transparent, with explainable sources
- Adaptable, letting you control what data is used in generation


Data Ingestion, Extensions & Workflows
Automate data ingestion from PDFs, spreadsheets, CRMs, ERPs, BI and cloud sources. Connect third party services and apps to keep all your knowledge synced and up to date.
With Extensions & Workflows, define how information is processed, synchronized, and enriched, ensuring your knowledge base stays continuously up to date.
Rational Knowledge in production

SBE-VARVIT
Technical drawings, datasheets and bills of materials connected in a knowledge graph. Search is conversational across sources, and each value is cited with its document and revision.


Google.org, LUISS DataLab, FIEG
An AI platform to combat disinformation, built with Geckosoft on Rational AI technology. Journalists explore large volumes of news through a knowledge graph and write with a Co-Editor that uses only the sources the newsroom selects, citing each one.


Pallex
Every delivery document, processed automatically with vision models. Signatures, stamps, handwritten notes and pallet returns are extracted into structured data that can be searched and analysed, anomalies are flagged before they turn into disputes, and the entire flow runs on Pallex's own infrastructure.

From data silos to → connected intelligence.
Visibility
See how knowledge connects and evolves as new data and relationships flow in.
Trust
Understand the reasoning behind every AI answer and action.
Productivity
Retrieve insights instantly from across your organization knowledge base and tools.