API and MCP
One interface to the knowledge behind the business.
Task-ready context or ranked evidence over HTTP, and the same capabilities to a compatible agent through MCP. Both deliver the same permission-aware understanding of the enterprise, grounded in the same sources and relationships — the identity is attached either way.
Percidian ACE
One governed understanding, one access decision, one set of sources.
- Context API
- Evidence API
- MCP
Context API
Request the context the task needs.
Compact, agent-ready context organized around a business task — for agents, copilots, workflow automation, and model prompts that need business-ready context.
POST /v1/context
- Request
- The task, the requesting identity, and the purpose of the work.
- Response
- Context, evidence, the permission decision, coverage, and source references — for that task and identity.
Evidence API
Keep control over retrieval and presentation.
Ranked, source-linked evidence for applications that assemble context themselves — custom retrieval workflows, research interfaces, evaluation systems, and applications that need granular evidence handling.
POST /v1/retrieve
- Entities and relationships
- The objects relevant to the task, and how they connect.
- Evidence and provenance
- The source behind each fact, with a path back to the original.
- Permissions, freshness, and coverage
- The access decision applied; how current and complete the result is.
MCP
Give agents a native way to understand the enterprise.
Expose approved Percidian context capabilities to a compatible agent through MCP. Users authenticate through the enterprise, and their permissions follow every request. The same agent can understand a client, prepare for a meeting, review a matter, or support a long-running workflow without receiving unrestricted access to the underlying systems.
Connect
Add the Percidian MCP endpoint to a compatible agent, assistant, or development environment.
Authenticate
Bind the session to the enterprise identity of the user or service.
Work with context
The agent requests only the business understanding the task needs.
Connect once. Govern centrally. Use across every compatible AI surface.
What a connected agent can request
- Understand a client or account
- Prepare a matter or deal briefing
- Find the evidence behind a business claim
- Retrieve the history of a decision
- Identify relevant obligations and deadlines
- Assemble context for a long-running workflow
Authentication and identity
The identity stays attached to the request.
User experiences use delegated enterprise identity; applications and workflows operate with approved service identities and defined scopes. If identity or permission cannot be established, the request receives no sensitive context.
User-delegated access
Service access
MCP authorization
Enterprise mapping
How each access decision is made and recorded: the request path.
Common build patterns
Use Percidian wherever AI needs to know the business.
The same interfaces, identity, and governed context, in the products teams build most often.
- In-product copilot
- Answers grounded in the user’s authorized customer, product, or account context.
- Enterprise assistant
- Institutional memory carried across conversations and AI applications.
- Workflow agent
- The history, relationships, and evidence a long-running agent needs to complete work.
- Research experience
- Structured, cited evidence for analysis and review.
- Meeting preparation
- The people, history, open issues, and decisions behind an upcoming interaction.
- Document generation
- Memos, reports, models, and presentations grounded in current enterprise context.
Bring your business into every AI decision
Give every approved AI system the context to do its best work.
Connect the knowledge your organization already has. Preserve what it learns. Apply the permissions it already trusts. Make that understanding available wherever people and agents work.