> ## Documentation Index
> Fetch the complete documentation index at: https://docs.aegra.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Why Aegra

> Self-host your AI agents with zero vendor lock-in, zero per-run costs, and full control over your data.

## The problem with managed agent platforms

Managed agent platforms charge per run, per node execution, and per seat. Costs scale linearly with usage and become unpredictable as your agents handle more traffic.

Beyond cost, there are structural trade-offs:

* **Data leaves your infrastructure.** Agent state, conversation history, and tool outputs are stored on third-party servers. You have no control over data residency or retention.
* **Tracing is locked to one vendor.** You can only send traces to the platform's own observability tool. Switching or adding a second backend is not supported.
* **No database access.** You cannot query, back up, or migrate the underlying database. The platform owns it.
* **Authentication is platform-managed.** You configure auth through a dashboard, not in your own code.

These are reasonable trade-offs for prototyping. They become problems at production scale.

## What Aegra gives you

* **Same SDK, zero code changes.** Use the LangGraph SDK you already know. Your graph code, client calls, and Agent Protocol frontends work without modification.
* **Your infrastructure, your data.** Run on your own servers with your own PostgreSQL database. Nothing leaves your network.
* **No per-run costs.** No seat fees, no per-execution charges, no node pricing. You pay only for your own compute and database.
* **Flexible authentication.** Write auth handlers in Python — JWT, OAuth, Firebase, or any custom scheme. Configured in `aegra.json`, not a dashboard.
* **Open tracing.** Fan-out traces via OpenTelemetry to Langfuse, Arize Phoenix, Jaeger, Honeycomb, Datadog, or any OTLP-compatible backend. Run multiple at once.
* **Agent Protocol compliant.** Works out of the box with Agent Chat UI, LangGraph Studio, and CopilotKit.
* **Apache 2.0 licensed.** Inspect, modify, and contribute. No proprietary dependencies.

## Cost comparison

|                         | LangSmith Deployments                                         | Aegra                                      |
| :---------------------- | :------------------------------------------------------------ | :----------------------------------------- |
| **Deploy agents**       | Local dev only (Free), paid cloud (Plus)                      | Free, unlimited                            |
| **Custom auth**         | Not available (Free), available (Plus)                        | Python handlers (JWT/OAuth/Firebase)       |
| **Scheduled cron jobs** | Not available (Free), available (Plus)                        | Built-in, free                             |
| **Self-hosted**         | Enterprise only (license key required)                        | Always (Apache 2.0)                        |
| **Own database**        | Managed only (Free/Plus), bring your own (Enterprise)         | Bring your own Postgres                    |
| **Tracing**             | LangSmith only                                                | Any OTLP backend (Langfuse, Phoenix, etc.) |
| **Data residency**      | LangChain cloud (Free/Plus), your infrastructure (Enterprise) | Your infrastructure                        |
| **SDK**                 | LangGraph SDK                                                 | Same LangGraph SDK                         |

*Based on [LangChain pricing](https://www.langchain.com/pricing) as of February 2026. An enterprise tier with self-hosting is also available at custom pricing.*

Aegra's infrastructure cost is whatever you pay for PostgreSQL and compute. A single \$5/month VPS can run the server, database, and multiple agents.

## What we support today

Aegra covers the core capabilities needed for production agent deployments: assistants, threads, streaming, human-in-the-loop, persistent state, key-value and semantic storage, auth, observability, and deployment to Docker, PaaS, or Kubernetes.

See the full breakdown on the [feature support](/feature-support) page.

<CardGroup cols={2}>
  <Card title="Streaming" icon="signal-stream" href="/guides/streaming">
    8 SSE stream modes with automatic reconnection.
  </Card>

  <Card title="Human-in-the-loop" icon="hand" href="/guides/human-in-the-loop">
    Approval gates and user intervention points.
  </Card>

  <Card title="Scheduled cron jobs" icon="clock" href="/guides/cron">
    Trigger runs on a schedule with timezone support and multi-instance safe claim.
  </Card>

  <Card title="Authentication" icon="lock" href="/guides/authentication">
    JWT, OAuth, Firebase, or custom auth handlers.
  </Card>

  <Card title="Observability" icon="chart-line" href="/guides/observability">
    Fan-out tracing to multiple OTLP backends.
  </Card>
</CardGroup>

## Next steps

<CardGroup cols={3}>
  <Card title="Quickstart" icon="rocket" href="/quickstart">
    Get a running server in under 5 minutes.
  </Card>

  <Card title="Migrate from LangSmith" icon="arrow-right-arrow-left" href="/migration">
    Step-by-step guide to switching from LangSmith Deployments.
  </Card>

  <Card title="Feature support" icon="list-check" href="/feature-support">
    Full matrix of supported, planned, and upcoming features.
  </Card>
</CardGroup>