Infrastructure that manages itself.
Infrastructure that manages itself. AI that knows how to help.
The Problem It Solves
Modern infrastructure is complex. Containers, microservices, multiple servers, certificates, configurations, dependencies. Managing it manually doesn't scale. Getting it wrong means outages, security gaps, and firefighting instead of building.
Worse: when something breaks at 3am, your team is searching Stack Overflow and guessing. The knowledge of how your infrastructure works lives in people's heads - or scattered across wikis nobody maintains.
You need infrastructure that deploys reliably, monitors itself, heals when it can, and - when it can't - tells AI exactly how to help.
What Odyssey Does
Odyssey is more than deployment orchestration. It's an AI-native operations platform - infrastructure where every component self-documents in a format AI can understand and act on.
Complete Lifecycle in One Manifest
Each application is defined in a single YAML manifest that covers its entire lifecycle: deployment configuration, health checks and monitoring, auto-remediation actions, troubleshooting knowledge, backup procedures, security requirements, client/agent installation, and AI-consumable documentation.
Not just "how to deploy" - a complete knowledge base for each application.
~100 Manifests, One Ecosystem
Odyssey manages AnnLighten alongside everything it depends on - databases (PostgreSQL, MariaDB), monitoring (Icinga), authentication (Keycloak), search (Elasticsearch, Meilisearch), vector stores (Milvus, Qdrant), messaging, and more.
All using the same schema. All supported by the same AI. All managed as one coherent platform.
AnnLighten isn't a special snowflake sitting on top - it's a first-class citizen in its own ecosystem.
AI-Native Operations
The manifest structure was designed for AI consumption. Your support AI doesn't search the internet or guess. It has authoritative, app-specific knowledge in a structured format. We already have AI trained for SL1/2 support.
Self-Healing Infrastructure
Health checks run continuously. When issues are detected, remediation actions execute automatically - restart services, reinitialise components, clear caches. The system fixes what it can and escalates what it can't.
Container and Bare-Metal
Most platforms handle containers OR bare-metal. Odyssey handles both - same schema, same AI support, unified management.
Core Capabilities
| Capability | What It Means |
|---|---|
| Complete lifecycle manifests | Deployment, health, remediation, docs - all in one place |
| AI-consumable structure | Every app self-documents for AI runtime support |
| ~100 app ecosystem | Databases, monitoring, auth, search, messaging - all managed identically |
| Auto-remediation | Self-healing actions when health checks fail |
| Multi-environment support | Same manifest, different servers via variable substitution |
| Certificate management | TLS certificates deployed and managed automatically |
| Visual health dashboards | HTML reports showing exactly what's running |
Who It's For
- Teams managing containerised applications who want AI-assisted operations
- Organisations running AnnLighten alongside its ecosystem
- Anyone tired of infrastructure knowledge living in people's heads
- Leaders who need self-healing infrastructure, not firefighting
The Difference
Most deployment tools stop at "containers are running." Odyssey continues to "AI can support this at 3am."
The manifest isn't just configuration - it's a knowledge graph. ~100 applications, all following the same schema, all queryable by AI, all managed as one platform.
Single schema, universal application. Most organisations have "special" deployment for their own software vs third-party. We don't. AnnLighten deploys exactly like PostgreSQL deploys exactly like Icinga deploys. That consistency is why the AI works.
Pragmatic, not clever. YAML, shell scripts, environment files. No exotic DSL, no complex orchestration framework. Anyone can read it. AI can act on it.
Odyssey was built for our own infrastructure - designed so AnnLighten could coexist with the open source tools it leverages. It reflects years of operational experience, not theoretical best practices.
