Skip to main content
OdinLabs
ProductInnovation SprintPricingDocsSecurity
  • Product
  • Innovation Sprint
  • Pricing
  • Docs
  • Security

No credit card required

Built in the Netherlands • Get started

  • Your data never leaves your infrastructure
  • ·Built in the Netherlands
  • ·EU jurisdiction by default
  • ·Open source core
OdinLabs

ODIN is AI you own. Deploy on your infrastructure, structure your organizational knowledge, and scale your team's capabilities. Built by Odin Labs in the Netherlands.

Product

  • How It Works
  • Use Cases
  • Pricing
  • Product

Company

  • About Us
  • Contact
  • Partners
  • Blog

Resources

  • Documentation
  • Integrations
  • Compare Tools
  • Security

Legal

  • Privacy Policy
  • Terms of Service
  • Cookie Policy

© 2026 Odin Labs Projects B.V. — Amsterdam, Netherlands. All rights reserved.

Designed for environments with audit, governance, and on-prem requirements.

BrainDBGOVERNED MEMORYagent:coding → code.commitreceipt #a8f2user:mitchell → policy.approvereceipt #b14cagent:brain → memory.writereceipt #c903agent:compass → decision.recordreceipt #d6e1user:dean → settings.updatereceipt #e7a8agent:sales → artifact.exportreceipt #f211
newSovereign Cloud Workspace

Code in your browser. On your infrastructure.

A per-tenant code-server + Odin Agent Runtime + workspace CLI, deployed on Hetzner-EU or the infrastructure you already control. Every commit, file write, and agent action gets a receipt.

Request enterprise accessSee it for solo + small teams
BuilderBuyerOperatorthree audiences · one surface
2.0Composition

Three pieces. One contract. Your boundary.

The workspace is composed from three primitives, each with its own audit surface. They work together because they share the same governance contract — not because they're glued at the seams.

primitive · code-server
container

Per-tenant code-server container. Mounted workspace volume survives container restart. Operator-set basicAuth + session cookie. Reachable at code.<slug>.staging.odin-labs.ai.

  • ship VS Code surface, browser-native
  • isolate Per-tenant network, no cross-tenant DNS
  • audit File-write events captured
uid1000 (coder)
networkper-tenant
primitive · odin-agent
CLI

Per-tenant Odin Agent Runtime. Skill manifest derived from the tenant's enabled hubs at provision time. 15 proven-core skills always loaded; opt-in beyond that.

  • deliberate Multi-round work-order planning
  • decide BrainDB write contract enforced
  • limit Per-tenant provider keys, per-tenant budget
version4.80.1
skills15 core + tenant
primitive · workspace CLI
HTTP

HTTP bridge between Command Center (uid 1001) and code-server (uid 1000). The workspace CLI handles cross-uid file writes so work-order updates from CC land cleanly inside the workspace.

  • bridge CC ↔ code-server I/O
  • contract sessionId ↔ workOrderId FK
  • verify Convergence ≤ 15s to CC kanban
authX-API-Key
cross-uidyes
3.0Sovereignty

Your boundary. Your data. Your jurisdiction.

On-prem on Hetzner-EU (Amsterdam) by default. Or deploy on your own Kubernetes / bare-metal / VPC. Data never leaves the boundary you choose. Per-tenant network isolation is a structural property — not a runtime check.

data-residency.live
EU
  • 🇪🇺EU sovereignty (Hetzner-NL default)yes
  • 🇳🇱Dutch jurisdictionyes
  • 🏢On-prem (your infrastructure)supported
  • 📜GDPR data-residency postureyes
  • 🔒Designed for HIPAA-eligibledesigned
  • 📋Designed for SOC 2 readinessdesigned

Per-tenant network isolation — structural, not runtime

  • networkEach tenant runs in its own Docker bridge network. Cross-tenant DNS resolution is a structural impossibility, not a firewall rule.
  • storagePer-tenant postgres database + per-tenant BrainDB. Append-only audit trail enforced at the storage layer.
  • envPer-tenant secrets and provider API keys. Operator-shared keys are explicit-opt-in only.
  • auditEvery cross-tenant attempt would itself emit an AuditEvent. Forensic by construction.

Doctrine source: rules 02 + 16 + 16c.

4.0Agentic

The workspace runs agents. The agents leave a paper trail.

The Odin Agent Runtime lives inside the workspace, not outside it. Your devs run /deliberate or /multi-track inside their VS Code session. Every spawned sub-agent is bounded by the tenant's budget + skill manifest + per-tenant provider keys.

terminal · odin-agent /deliberate
+4-0bash
1 $ odin-agent /deliberate "add OAuth2 to the customer portal"
2
3 Phase 0 → tools detected: codex T5, factory T1, droid T2
4 Phase 0.5 → subject pinned, cycle-context written
5 Phase 1 → cognitive council confidence 78%
6 Phase 2 → 4 WOs drafted, validator GREEN
7+ WO-01 → src/auth/oauth2.ts (new)
8+ WO-02 → src/auth/oauth2.test.ts (new)
9+ WO-03 → src/routes/auth.ts (modified)
10+ WO-04 → docs/auth.md (new)
11 Phase 3 → execution awaiting your approval
every phase emits AuditEvent · cycle id del-add-oauth2-1780531000

Where the runtime lives

  • →Inside the workspace — your VS Code session has the CLI on its PATH. No remote API to call.
  • →Per-tenant manifest — skill set scoped to the hubs your tenant enabled at provision.
  • →Per-tenant providers — Anthropic / OpenAI / OpenRouter / local MLX keys come from your tenant env, not a shared pool.
  • →Per-tenant budget — daily + per-session caps stop runaway spend by construction.
  • →Tenant-scoped audit — every spawn + every prompt + every tool call lands in your BrainDB.
Explore Odin Agent Runtime→
5.0Audited

Every action gets a receipt.

The workspace doesn't bolt on audit. It emits AuditEvent at every governance boundary — file writes, agent spawns, model calls, BrainDB writes, policy approvals. The trail is queryable by actor, action, tenant, timestamp.

workspace · audit-trail · last 8 events
decisions
  • 14:23:14user:operator→workspace.open#a8f2

    “VS Code session opened on code.demo-3.staging.odin-labs.ai”

  • 14:24:01agent:coding→deliberate.phase-1#b14c

    “Cognitive council confidence 78% on OAuth2 ambition”

  • 14:25:30agent:coding→code.edit#c903

    “src/auth/oauth2.ts created (124 lines)”

  • 14:25:31agent:brain→memory.write#d6e1

    “Decision recorded: chose authlib over passport (rationale 18 chars)”

  • 14:28:42user:operator→policy.approve#e7a8

    “Approved auth/oauth2 PR via Control Tower kanban”

  • 14:29:01agent:coding→code.commit#f211

    “Committed feat/oauth2 to workspace git remote”

  • 14:31:18agent:brain→memory.query#a45d

    “Semantic recall: prior authlib decision surfaced for cross-team review”

  • 14:32:00user:operator→workspace.close#b8f9

    “Session ended, transcript preserved in BrainDB”

forensically queryable · per-tenant isolated · append-only

Your workspace. Your boundary. Your paper trail.

Pick the path that fits your environment.

Request enterprise accessSee it for solo + small teams

Related products

Workspace

Sovereign Cloud Workspace

Code in your browser. On your infrastructure. Governed by construction.

Learn more →
Engines

Hub Engines

Six domain-isolated engines that ship code, decisions, content.

Learn more →
Memory

BrainDB

Governed memory. Append-only audit. Semantic recall over the why.

Learn more →
Ops

Control Tower

Operator dashboard for your tenant fleet.

Learn more →
Assistant

Odin Assistant

Voice + chat assistant for the workspace. Multi-provider. Audited.

Learn more →