← Back to the start page
KritiGraph for Organisations
KritiGraph answers questions from your own document base — with AI whose accesses are mediated under control and recorded traceably.
The problem KritiGraph addresses
Organisations holding confidential knowledge face a dilemma: AI-supported answers from their own documents are useful, but common AI services process content outside the organisation's infrastructure, and individual accesses are hard to evidence afterwards.
KritiGraph addresses both points: processing takes place where the deployment model prescribes it, and governance decisions about model and knowledge accesses are kept as durable records that non-technical staff can read.
Who it is for
Built for organisations where the place of data processing and the evidencability of accesses matter: law firms and holders of professional secrecy, public administration, operators of critical infrastructure, and companies with internal confidentiality tiers.
What “local” and “controlled” mean
Local means: documents, indexes and language models run in your own infrastructure; which claims apply in which deployment model is stated per claim (deployment-model scoping on the evidence page).
Controlled means: accesses to models and protected knowledge go through a dedicated governance kernel that checks requests before execution and records the decision. The technology page describes how this is built.
What is planned
In development, not yet a product commitment: rolling independently maintained knowledge bases back to an earlier eligible version, and connecting the public claims to the central assurance matrix so their assurance status is derived automatically.
Evidence available today
The following claims are projected onto this page from the machine-readable claim source; their current state is kept on the evidence page.
Knowledge stays separated — mandates do not intermix.
Your domain knowledge can be maintained independently of the AI models used.