Turning fragmented data into foresight and action.
Accentrust builds AI-native infrastructure that helps enterprises turn fragmented data into governed context, reliable workflows, and measurable decisions.
Explore our productsProducts & Platforms
Figena
Finance and operations, run on AI-native infrastructure
Cilila AI
A governed AI platform for real business work
Figena+ your applications
- FabricOne reliable picture of the whole organization
- GuardEvery access, action, and approval within policy
- StudioAutomation shaped around how each team works
- SignalsDecision-ready guidance from everyday operations
Models & inference
Research behind the platform
We publish how AI agents should be given access, kept within policy and held to account, as open papers and an open protocol.
Authorization Revocation for Long-Running AI Agents: Root-Scoped Quiescence under Delegation and Asynchronous Execution
Runtime Authorization for Resources Acquired by AI Agents
Versioned Transitive Dependency-Closure Binding and Operation-Time Effect Governance for Agent Skills: ClosureBound
- 24
- papers and working papers published
- Open preprints, free for anyone to read
- An open protocol for AI tool access, with its code public.
View on GitHub
Articles
Practical notes on putting AI to work in a business: where to start, how to evaluate it, and how to keep it within policy.
How Accentrust Powers Figena With Five Platform CapabilitiesUnderstand how Figena brings business work into one application and how Accentrust's five capabilities support context, intelligence, control, workflows, and insight.
Choosing the First Workflow for an Enterprise AI PilotChoose an AI pilot with a clear task, suitable inputs, a named reviewer, and results people can check. Define its boundary and decide when to continue, narrow, or stop.
What Enterprise Buyers Should Ask About an AI SystemEvaluate an AI purchase through specific workflows and evidence. Ask about data paths, permissions, review, recovery, retention, and exit before accepting the proposed scope.