UmamiAI • Agentic Platform•

Build agents that execute workflows—safely and at scale

A production-grade scaffold: multi-tenant orgs enforced by Postgres RLS, versioned agents, knowledge (RAG), approvals, API keys, webhooks, and auditable runs. Extend it into your full agent OS.

Read setup docs

Multi-tenant by default

RLS

Every table is scoped by org. Policies enforce isolation and role-based access without app-layer hacks.

Versioned agents

v1+

Publish prompt + model changes as versions. Run traces remain reproducible and auditable.

API keys & webhooks

v0.6

Issue scoped keys per org and receive signed webhooks for downstream automation and observability.

Approvals & policies

gates

Require approvals for risky tool actions and enforce budgets + allowlists at the org layer.

Developer quickstart

Integrate UmamiAI in minutes: create an API key, call the Platform API, and optionally subscribe to signed webhooks.

# 1) Create an API key in Console → Org → API Keys
export UMAMIAI_BASE_URL="https://YOUR_DOMAIN"
export UMAMIAI_API_KEY="umami_..."

# 2) Start an agent run
RUN_ID=$(curl -sS -X POST "$UMAMIAI_BASE_URL/api/v1/agent-runs" \
  -H "Authorization: Bearer $UMAMIAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d "{\"agent_id\":\"<AGENT_ID>\",\"input\":\"Summarize last week\'s sales and flag anomalies.\"}" \
  | jq -r .id)

# 3) Fetch run status/output
curl -sS "$UMAMIAI_BASE_URL/api/v1/agent-runs/$RUN_ID" \
  -H "Authorization: Bearer $UMAMIAI_API_KEY" | jq

# 4) (Optional) Webhooks
# Configure Console → Org → Webhooks and verify X-Umami-Signature on delivery.

View Platform API docs

What you get

Control plane UI (Next.js) with secure cookie sessions
Data plane (Supabase Postgres) with RLS policies and queues
Worker loop for embeddings + run traces + tool calling