Run agents
Use native caveauAI agents for answers, research, drafting, support triage, and tool-assisted work under workspace permissions.
- Cited RAG from approved corpora
- Tool calls through governed MCP
- Run history and observability
Run, build, and govern
AI Agents for governed business work
caveauAI Agents turns your documents, tools, and workflows into auditable AI systems: cited answers, human approvals, tenant-scoped tools, and managed setup when you want the first useful agent live without a long platform project.
RAGCited knowledge answers
MCPScoped tools and audit trails
EUSovereign hosting path
Four ways in
Pick the agent path that matches the job.
The customer-facing promise is simple: start with a governed agent, then choose whether to build visually or with code as the work gets more serious.
Build visually
Map workflows and LLM flows in a visual builder, then bring the useful parts back into caveauAI governance.
- Node-RED standard for workflows
- Flowise Community by request
- n8n dedicated runtime by request
Build with code
Use LangChain and LangGraph when the agent needs state, branching, tests, and real software engineering discipline.
- Connection kit for every workspace
- Hosted LangGraph premium workspace
- Scoped MCP and model gateway access
Author skills
Package what an agent knows how to do into a governed, versioned skill any tool or MCP host can load — lint it for quality first.
- Governed, versioned skill packages
- Built-in quality linting
- Exports to any Agent-Skills host
Use-case chooser
What are you trying to automate?
A quick recommendation layer helps a buyer understand the stack without learning the names of every builder first.
Recommended: caveauAI Agents
Start with a governed RAG/tool agent.
Best when the work depends on your private documents and must stay cited, logged, and permission-aware. Add tools only after the answer flow is trusted.
- Use private corpora plus approved public packages.
- Keep tool permissions narrow and observable.
- Book setup if the workflow touches customers or operations.
Read-only public demo
Ask public knowledge before you book.
This demo searches public Knowledge Exchange packages only. It cannot run tools, trigger workflows, send email, or access customer workspaces.
Answer
Ready
Choose a package and ask a question. Answers appear here with source snippets.
Customer packages
Setup first, tooling second.
The product is not another empty builder. It is a route to a useful agent with the right knowledge boundary, approval model, and runtime choice.
caveauAI Agents
Native governed agents for RAG, tools, drafting, support, and repeatable work inside the portal.
Node-RED
Apache-2.0 visual workflows with caveauAI nodes for search, ingestion, and governed tool calls.
Flowise Community
Dedicated visual LLM flow builder for teams that want graph-style experimentation before governance.
LangGraph
Code-first stateful agents with a connection kit for all customers and hosted workspace by request.
Skill Maker
Author, lint, and version reusable skills. 64 published skills across 5 client workspaces so far.
Book setup
Bring one real workflow. We will map the agent around it.
Best first setup: one knowledge set, one business outcome, one approval boundary. The rest can grow after the first agent proves useful.