v1.2 shippedRBAC + OAuth live
Backed by retrieval, not vibes

The knowledge OS
for the agentic era.

RAGNES ingests every doc, thread, and ticket your company has ever written, structures it into an executable skills file, and hands it to your team — and their agents — with citations on every answer.

SOC 2-ready Multi-tenant
ragnes ▸ chat.stream● live
// stream a sourced answer
await fetch('/api/chat/stream', {
method: 'POST',
body: JSON.stringify({ message, session_id })
})
// → meta { citations } → delta … → done
assistant
Refunds are handled by Support within 48h of the ticket reaching the queue [#1]. For orders over $500, escalate to Finance [#2]
#1 · refund-policy.md#2 · finance-sla.pdf

Connect what your team already uses

SlackHubSpotGmailNotionSalesforceGoogle DriveGitHubZendeskJiraLinearZapiern8nSlackHubSpotGmailNotionSalesforceGoogle DriveGitHubZendeskJiraLinearZapiern8n

p95 latency

182ms

streaming chat

sources

12 kinds

docs · Slack · email · CSV

citations

100%

every answer, sourced

roles

3-tier

admin · member · viewer

The product

A knowledge engine with the ergonomics of a codebase.

Six primitives. One integrated surface. Every capability is API-first, workspace-scoped, and cited by default.

Ingest

Every doc, PDF, thread, ticket.

PDFs, DOCX, CSVs, Slack exports, .mbox archives, plain text. Parsed, deduped, and embedded in seconds.

Structure

Auto-classified. Actually.

Claude Sonnet sorts every atom into Process, Policy, Decision Rule, Role, or FAQ — with a visual map you can navigate.

Retrieve

Answers with citations. Always.

Streamed responses over SSE with inline [#n] source chips. If we don't know, we say so.

Export

Skills File for your agents.

One JSON per department. Zapier, n8n, Cursor, or your own agent framework can consume it directly.

Govern

Roles, invites, workspaces.

Admin / Member / Viewer. Multi-tenant isolation. Invite by email or shareable link. Audit built in.

API

Agent-native from day one.

Per-key rate limits, IP whitelists, p95 latency, rotation with grace windows. Ship agents to prod, not toys.

How it works

From scattered docs to shipping answers in three moves.

01

Connect

OAuth into Slack, HubSpot, Gmail — or drop docs directly. RAGNES ingests, chunks, and embeds every source in the background.

02

Structure

Sonnet 4.6 classifies each atom into a canonical schema. You get a searchable knowledge graph, not a folder of PDFs.

03

Deploy

Query from the web app, hit the streaming API, or hand the skills-file JSON to your agent. Answers ship with citations.

For developers

Agent-native, out of the box.

Per-key rate limits, IP whitelisting, key rotation with grace windows, p95 latency analytics, and a batched analytics writer. Ship agents that talk to your knowledge — not scripts that scrape your Notion.

  • Streaming SSE chat with citations on every frame
  • Workspace-scoped bearer keys · one row, one bcrypt hash
  • Redis-backed token bucket + in-memory LRU fallback
  • Debounced re-embedding on knowledge edits
ragnes ▸ agent.query
curl -X POST $RAGNES_URL/api/agent/query \
  -H "Authorization: Bearer $RAGNES_KEY" \
  -H "Content-Type: application/json" \
  -d '{"question": "How do we handle refunds?", "top_k": 5}'
200 OK · latency_ms: 182
results: [ 5 documents · sorted by cosine · cited ]

Pricing

Free while you build. Fair when you ship.

Free

$0

Get started

  • 1 workspace
  • 3 seats
  • 500 knowledge entries
  • Community support

Team

Popular

$29

Per seat / month

  • Unlimited workspaces
  • Slack / Gmail / HubSpot OAuth
  • Rate-limit self-healing
  • Priority support

Scale

Custom

Enterprise

  • Atlas Vector Search
  • SSO + SCIM
  • SLA + dedicated support
  • Custom integrations

Stop asking Slack. Start asking RAGNES.

Free for teams under 3. No credit card. Bring your Claude / OpenAI key or use ours.