Catalog approved, internal, third-party, and discovered agents with owners, purpose, providers, access, and lifecycle.
Use approved agents in chat and project workspaces with governed files, knowledge, connectors, skills, and optional memory.
Connect access decisions, security events, health, usage, cost, and audit history to the agent and the people responsible for it.
Teams build and buy agents across frameworks, SaaS tools, automation platforms, and internal services.
Pristan creates a governed record for each agent, whether it comes from a manifest, API, integration, or manual review.
Ownership, lifecycle, access, evidence, and cost live with the agent instead of across separate spreadsheets and tools.
Create catalog records for custom frameworks, internal services, third-party tools, and provider-backed agents.
Keep each agent accountable from request and approval through production, review, and retirement.
Keep audit evidence, security signals, usage, and spend connected to the agent record.
An employee asks the HR bot about parental leave. It answers "4 weeks." Your policy is 16. Now someone's planning their life around bad info.
Assign approved knowledge bases or ready personal and project files to the agent workflows that should use them.
When retrieval contributes context, users can inspect the citations and sources returned with the answer.
Upload supported documents for extraction, scanning, chunking, and indexing under administrator-defined storage and processing rules.
Use built-in pgvector or configure a supported external vector adapter for the knowledge base.
When enabled, users can manage private files and teams can work with shared project descriptions, files, and conversations.
Capabilities are policy-controlled and appear only when administrators enable and configure them.
Stream responses, manage conversation history, attach allowed files, inspect citations, rate answers, and apply configured DLP and memory behavior.
Share project descriptions, files, conversations, and collaborator permissions. Copy eligible personal chats into a project without copying private files, connector references, artifacts, ratings, or memory.
Install exact connector versions, control agent access, expose approved sources in chat, and use encrypted per-user credentials when a connector supports user-owned authentication.
Import externally created skill packages, validate and version them, publish reviewed versions, and assign them to the agents allowed to use them.
Create and refine Markdown artifacts when enabled for an agent, and dictate prompts when an administrator has configured an approved transcription provider.
When memory is enabled, users can review, edit, import, export, or forget global and agent-specific memories within administrator-defined policy.
Ownership, approvals, access decisions, reviews, and evidence stay connected to the agent record.
Bring DLP findings, prompt-risk context, and control evidence from the tools already in your stack.
Query usage, cost, risk, and lifecycle status in the same system that owns the catalog.
Permissioned compliance review of conversation content and metadata.
Detect and record prompt-injection signals, then apply the configured action and agent policy.
Use natural-language administrative queries and typed actions when Platform AI is configured, with the caller's permissions still enforced.
Choose a private self-paced workspace or a guided walkthrough. We will focus the evaluation on the agent catalog, everyday work, access controls, security evidence, connectors, and deployment questions that matter to your team.