---
title: Deploy AnythingLLM with AI on Blossom
url: "/deploy/anythingllm/"
canonical_url: https://docs.blossom88.com/deploy/anythingllm/
markdown_url: https://docs.blossom88.com/deploy/anythingllm.md
summary: Deploy AnythingLLM with AI on Blossom. Build a private document assistant
  with saved workspaces, cited answers, and persistent storage.
categories:
- deployment
- ai
---

# Deploy AnythingLLM with AI on Blossom

[AnythingLLM](https://anythingllm.com/) lets you chat with your documents in
private workspaces. Upload reference material, connect a model, and ask
questions with links back to the source.

Its Node.js server and React interface keep workspace records in SQLite and
can use LanceDB for local vectors. Browse the
[source on GitHub](https://github.com/Mintplex-Labs/anything-llm).

Run it as a [Persistent Service](/docs/apps/persistent-services/)
so documents, chat history, and vectors stay with the deployment.

[Install the Blossom CLI and skill](/learn/getting-started/install/)
on your agent's machine and choose a Blossom Cloud Location when you log in.
You can also use
[Blossom BYOC (bring your own cloud)](/learn/getting-started/choose-your-cloud/#blossom-byoc).

## Ask Your Agent

> **Ask your AI agent:** Deploy AnythingLLM on Blossom as a private document assistant. Help me connect my model provider and ask a question about a file.

> Follow this guide: https://docs.blossom88.com/deploy/anythingllm/



Your agent checks the current
[Docker requirements](https://docs.anythingllm.com/installation-docker/overview),
chooses a compatible maintained image, and saves the configuration. Supply
your model provider's API key through a private environment file. This setup
uses a hosted model and embeddings; it doesn't run a local model or need a GPU.

| Part | Purpose |
|---|---|
| AnythingLLM Service | Workspace interface, document processing, and chat |
| Persistent Volume | SQLite, documents, local vectors, and saved settings |
| Your model provider | Generates embeddings and answers |

Keep the app behind **Blossom Account access** and configure its own password.
Preserve its signing keys and encryption secrets through updates. Documents
and questions sent to a hosted provider are subject to that provider's data
policy.

## Ask About Your Documents

> **Ask your AI agent:** Create a workspace, add this document, and help me ask a question using only its contents.

> Follow this guide: https://docs.blossom88.com/deploy/anythingllm/



Upload a small reference file and add it to the workspace. Choose **Query**
mode when you want answers grounded in the documents. Ask a question with an
answer you can check, then open **Sources** to see the supporting file.

![AnythingLLM answering a question from an uploaded document and showing its source.](/assets/images/guides/anythingllm/grounded-chat.png)

<details class="optional-reading" markdown="1">
<summary>Commands behind the deployment</summary>

Create a prebuilt-image Environment and a Persistent Service:

    blossom app:create --data app.yml
    blossom environment:create --data environment.yml
    blossom env_var:push --environment environment-100 --file .env.production
    blossom service:create --data anythingllm-service.yml
    blossom environment:deploy --id environment-100 --follow

Keep `STORAGE_DIR` at `/app/server/storage` and mount the Volume there. Use the
image's supported startup entrypoint and HTTP check at `/api/ping` on port
`3001`. Preserve the image's non-root user.

Service storage:

```yaml
storage:
  enabled: true
  mount_path: /app/server/storage
  size_gib: 10
```

AnythingLLM also writes settings to `/app/server/.env`. Keep that file on the
Volume: initialize it from the image only when absent, then link it back to
`/app/server/.env` before the supported entrypoint starts. Save the startup
wrapper with the deployment. Don't overwrite saved settings on every start.

Configure the language model and embedding provider separately. Set stable
`JWT_SECRET`, `SIG_KEY`, and `SIG_SALT`, a private `AUTH_TOKEN`, and disable
built-in local inference when using a hosted model. Keep API keys private.
Check the selected release's current configuration names.

</details>

## Keep Your Workspace Through Updates

> **Ask your AI agent:** Update AnythingLLM and check that my workspace, chat history, and uploaded document still work.

> Follow this guide: https://docs.blossom88.com/deploy/anythingllm/



Keep the same Volume, provider configuration, and secrets. After replacement,
sign in again, reopen the earlier conversation, and ask another question about
the saved document. Check its source citation too.

This SQLite and local-vector setup runs **one Container**. A replacement stops
the previous writer before starting the next one. Back up the Volume and test
a restore before relying on it for important material.

## What You Get

- A private document assistant with saved workspaces
- Answers grounded in uploaded reference material, with source citations
- Chat history, documents, and vectors kept across Container replacement
- Your choice of a supported model and embedding provider
