Documentation
Key Concepts
A quick reference for the terms used throughout LithicBlue and the rest of these docs.
Glossary
- Project
- The container for one piece of work: its source documents, scope, analysis and deliverables, and the people who can access it.
- Sources
- The documents you put into a project: board packs, interview notes, financials, prior reports. LithicBlue reads them to build the project knowledge base.
- Deliverables
- The outputs you produce for the client (presentation decks, PDFs and Word documents), which are reviewed against your sources rather than feeding the evidence base.
- Project knowledge base
- What LithicBlue builds from your project's own sources so the AI can answer questions, run analysis and verify claims with the whole project in context. Distinct from the external knowledge bases (Wikipedia, ArXiv, World Bank) the research agent can also draw on.
- Ontological map
- LithicBlue’s proprietary, professional-services ontology: the structured map it builds of the people, organisations, topics and key facts across a project’s sources and how they connect. It is what lets LithicBlue reason across the whole project rather than one document at a time, and trace every claim back to where it came from. The ontological map is what gives the project knowledge base its structure.
- Claims & verification
- A claim is a factual assertion in a deliverable. Verification checks each one against your sources and gives it a verdict (verified, contradicted or unverified) anchored to the exact passage.
- Scope
- What the project set out to answer, broken into sections and the requirements within them. LithicBlue tracks how well your sources fulfil each requirement.
- Data gaps
- The open questions a project can’t yet answer from the documents provided: the evidence you still need to gather.
- Contradictions
- Places where claims drawn from different sources disagree. LithicBlue surfaces them for you to validate or override rather than silently picking a winner.
- Feedback agents
- A panel of expert AI agents (visualization, writing, industry, structure, plus your own feedback personas) that critique a deliverable from different perspectives.
How they fit together
You bring sources into a project; LithicBlue reads them into a knowledge base, structured by its ontological map; you define the scope and watch the data gaps close as evidence arrives; it flags contradictions for you to resolve; and the deliverables you produce carry claims that are verified and traceable back to the source.