¶ Knowledge Framework
The Knowledge Framework enables structured, governed access to knowledge artifacts, embeddings, and document sources across the Fuse Platform.
It allows developers to integrate external knowledge systems, manage dynamic knowledge artifacts, and make information accessible across workflows, AI operations, human tasks, and runtime executions — all while preserving security, auditability, and policy enforcement.
¶ Overview
Fuse treats knowledge artifacts as first-class, composable runtime resources — available to humans, AI agents, workflows, and external systems alike.
The Knowledge Framework empowers the platform to:
- Ingest and manage knowledge from documents, wikis, APIs, pipelines, or external systems
- Represent knowledge in multiple forms — pages, nodes, links, embeddings, or real-time providers
- Provide real-time or persistent access to structured knowledge
- Govern usage through token-scoped, role-based, and context-aware policies
- Compose and inject knowledge into runtime flows, prompts, and decisions
- Extend Fuse’s native Resources domain with dynamic and external content providers
¶ Key Capabilities
| Capability | Description |
|---|---|
 Knowledge Artifact Management |
Govern and transform knowledge artifacts: documents, nodes, embeddings, pages |
 Resource Integration |
Connect internal files, URLs, workflows, or external APIs as structured resources |
 Dynamic Retrieval and Indexing |
Enable runtime or persistent knowledge access, with optional caching or vector search |
 Provider Extensibility |
Register external systems (e.g., Wiki.js, search APIs) as live knowledge providers |
 Vector Database Support |
Power semantic search and context injection via embeddings and vector stores |
 Security and Policy Enforcement |
Apply identity-aware access rules, session context, and domain-scoped permissions |
All capabilities are built to operate securely under Fuse’s unified access model, with explainability and traceability by default.
¶ How It Interacts with Fuse Domains
| Domain | Integration Role |
|---|---|
| Intelligence | Enables retrieval-augmented generation (RAG), prompt building, and semantic enrichment |
| Workflow | Injects knowledge into task steps, form fields, validations, and approval flows |
| UI | Powers context panels, help widgets, dashboards, and data views with linked knowledge |
| Security | Enforces scoped, token-based, and resource-specific access to knowledge artifacts |
| Storage | References files, blobs, workflow outputs, and external knowledge locations |
¶ Common Use Cases
Supplying AI assistants with verified, scoped knowledge during conversation or action
Injecting domain-specific knowledge into approval or task steps- Dynamically retrieving and composing output from live or hybrid knowledge sources
Reusing previous workflow outcomes as versioned knowledge artifacts
Enhancing semantic search and assistant performance with embedded document knowledge
¶ Extensibility
The Knowledge Framework is designed for composability and extension.
You can extend it by:
- Registering new knowledge providers with custom capabilities
- Integrating external vector databases (e.g., FAISS, Qdrant, Pinecone)
- Writing pipelines that output dynamic knowledge resources
- Defining custom ontologies and node/link types
- Configuring embedding services and re-indexing workflows
- Implementing scoped policies for versioning, trust scoring, and audit logging
¶ Developer Resources
To explore technical implementation:
These pages provide structured guidance for working with each part of the Knowledge Framework.
Next: Messaging Framework — Learn how to trigger secure, governed messages and notifications across domains and workflows.