AI Bedrock support (#239)

* AI Bedrock support

* formatting fixes

Author: danciaclara

docs/self-hosting/govern/environment-variables.md

@@ -198,18 +198,19 @@ Plane AI uses a separate PostgreSQL database. Create a new database (e.g. `plane
 
 Plane AI supports multiple LLM providers. Configure one or more by adding their API keys.
 
-| Variable                   | Description                                                     | Required |
-| -------------------------- | --------------------------------------------------------------- | -------- |
-| **OPENAI_API_KEY**         | API key for OpenAI models                                       | Optional |
-| **CLAUDE_API_KEY**         | API key for Anthropic models                                    | Optional |
-| **GROQ_API_KEY**           | API key for speech-to-text features                             | Optional |
-| **CUSTOM_LLM_ENABLED**     | Set to `true` to use a custom LLM with an OpenAI-compatible API | Optional |
-| **CUSTOM_LLM_MODEL_KEY**   | Identifier key for the custom model                             | Optional |
-| **CUSTOM_LLM_BASE_URL**    | Base URL of the custom model's OpenAI-compatible endpoint       | Optional |
-| **CUSTOM_LLM_API_KEY**     | API key for the custom endpoint                                 | Optional |
-| **CUSTOM_LLM_NAME**        | Display name for the custom model                               | Optional |
-| **CUSTOM_LLM_DESCRIPTION** | Description of the custom model                                 | Optional |
-| **CUSTOM_LLM_MAX_TOKENS**  | Maximum token limit for the custom model                        | Optional |
+| Variable                  | Description                                                                                                                                            | Required |
+| ------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------ | -------- |
+| **OPENAI_API_KEY**        | API key for OpenAI models                                                                                                                              | Optional |
+| **CLAUDE_API_KEY**        | API key for Anthropic models                                                                                                                           | Optional |
+| **GROQ_API_KEY**          | API key for speech-to-text features                                                                                                                    | Optional |
+| **CUSTOM_LLM_ENABLED**    | Set to `true` to enable a custom LLM. Supports OpenAI-compatible endpoints and AWS Bedrock.                                                            | Optional |
+| **CUSTOM_LLM_PROVIDER**   | Backend provider for the custom model. Accepted values: `openai` (default), `bedrock`.                                                                 | Optional |
+| **CUSTOM_LLM_MODEL_KEY**  | Identifier key for the custom model (e.g. a model ID or name).                                                                                         | Optional |
+| **CUSTOM_LLM_BASE_URL**   | Base URL of the custom model's OpenAI-compatible endpoint. Required when `CUSTOM_LLM_PROVIDER=openai`.                                                 | Optional |
+| **CUSTOM_LLM_API_KEY**    | API key for authenticating with the custom endpoint. Required for `openai` provider; used as the AWS access key ID when `CUSTOM_LLM_PROVIDER=bedrock`. | Optional |
+| **CUSTOM_LLM_AWS_REGION** | AWS region for the Bedrock model (e.g. `us-east-1`). Required when `CUSTOM_LLM_PROVIDER=bedrock`.                                                      | Optional |
+| **CUSTOM_LLM_NAME**       | Display name for the custom model shown in the UI. Defaults to `Custom LLM`.                                                                           | Optional |
+| **CUSTOM_LLM_MAX_TOKENS** | Maximum token limit for the custom model. Defaults to `128000`.                                                                                        | Optional |
 
 #### Provider base URLs
 

docs/self-hosting/govern/plane-ai.md

@@ -43,10 +43,15 @@ You can provide API keys for both OpenAI and Anthropic, making all models availa
 
 #### Custom models (self-hosted or third-party)
 
-Plane AI works with any model exposed through an OpenAI-compatible API, including models served by Ollama, Groq, Cerebras, and similar runtimes. You can configure one custom model alongside your public provider keys.
+Plane AI supports custom models through two backends:
 
-:::warning
-For reliable performance across all Plane AI features, use a custom model with at least 100 billion parameters. Larger models produce better results.
+- **OpenAI-compatible endpoint** — any model exposed via an OpenAI-compatible API, including models served by Ollama, Groq, Cerebras, and similar runtimes.
+- **AWS Bedrock** — models accessed directly through Amazon Bedrock using your AWS credentials.
+
+One custom model can be configured alongside your public provider keys.
+
+::: warning
+The custom model should have at least 100 billion parameters for all Plane AI features to work reliably. Larger, more capable models yield better results.
 :::
 
 ### Embedding models
@@ -114,20 +119,22 @@ CLAUDE_API_KEY=xxxxxxxxxxxxxxxx
 
 ### Custom model
 
-Use this for self-hosted models or third-party OpenAI-compatible endpoints.
-
 ```bash
 CUSTOM_LLM_ENABLED=true
+CUSTOM_LLM_PROVIDER=openai  # or 'bedrock'
 CUSTOM_LLM_MODEL_KEY=your-model-key
-CUSTOM_LLM_BASE_URL=http://your-endpoint/v1
 CUSTOM_LLM_API_KEY=your-api-key
 CUSTOM_LLM_NAME=Your Model Name
-CUSTOM_LLM_DESCRIPTION="Optional description"
 CUSTOM_LLM_MAX_TOKENS=128000
 ```
 
-:::info
-The custom endpoint must expose an OpenAI-compatible API matching OpenAI's request and response format.
+**Additional required variables by provider:**
+
+- **OpenAI-compatible** (`openai`): `CUSTOM_LLM_BASE_URL`
+- **AWS Bedrock** (`bedrock`): `CUSTOM_LLM_AWS_REGION`
+
+::: warning
+For Bedrock, the IAM user must have `bedrock:InvokeModel` permission on the target model.
 :::
 
 ### Speech-to-text (optional)