Platform Beta - Docker Compose Setup

• Open-Responses: https://github.com/masaic-ai-platform/open-responses
• Platform-UI: https://github.com/masaic-ai-platform/platform-ui

Prerequisites

• Docker and Docker Compose installed
• Required ports must be available on your host system:
  • 6644 (Open Responses API)
  • 6645 (Platform UI)
• See Cleanup Commands at the end if you need to remove existing containers/images.

Deployment Modes

Mode	Profile(s)	Services	Use Case
1. Default	default	open-responses, platform-ui	Basic local testing
2. Qdrant	qdrant	open-responses-vector, qdrant, platform-ui	Vector search with Qdrant
3. MongoDB	mongo	open-responses-mongo, mongo, platform-ui	chat messages persistence with MongoDB
4. Full	full	open-responses-full, mongo, qdrant, platform-ui	Full stack (vector + chat messages persistence)
5. External	external	open-responses-external, platform-ui	Use external Qdrant/MongoDB
6. Hybrid	mongo,external or qdrant,external	Mix of internal/external DBs	Custom hybrid setup

---

How to Run Each Deployment Mode

1. Default (Basic)

docker-compose up

2. Qdrant (Vector Search)

docker-compose --profile qdrant up

3. MongoDB (Persistence)

docker-compose --profile mongo up

4. Full (Qdrant + MongoDB)

docker-compose --profile full up

5. External (External Qdrant/MongoDB)

docker-compose --profile external up

6. Hybrid (Mix Internal/External)

• External Qdrant, Internal MongoDB:

  docker-compose --profile mongo --profile external up

• Internal Qdrant, External MongoDB:

  docker-compose --profile qdrant --profile external up

To bring down a mode, use the same profile(s) with down:

docker-compose down
# or, for a specific mode:
docker-compose --profile full down
# ...etc.

---

📝 Feedback

We'd love to hear your feedback! Please take a moment to fill out our feedback form - it takes less than 4 minutes.

---

Use Cases

1. Add Model: Add your own model with built-in test

Steps

1. Add model base url, example: if /chat/completions is available at http://example.com/v1/chat/completions then add http://example.com/v1
2. Built-in model connectivity test runs.
3. Save model and use it to build use cases.

2. Ecommerce: Product Search and Add to Cart in One Shot (Grok4)

System Prompt:

1. Use all birds tool for product search and update cart.
2. Return image URL in a proper markdown format.
3. Whenever you update cart, share checkout link also.

MCP Server:

https://allbirds.com/api/mcp

Sample Query:

Add men's black runner jet black, black sole of size 9 in the cart.

---

3. File Search Tool, In-Built RAG

System Prompt:

Use file search tool to frame your answers

On File Search Tool Modal:

1. Create vector store, using '+' icon.
2. Select embedding model.
3. Upload file to store.
4. Select store and file and generate embeddings using 'Associate File CTA'
5. Select vector store and save the tool

Sample Query: Ask relevant query which will invoke file search tool.

---

4. Generating Mock Tools Using Masaic Mocky

Once a mock is generated, create an MCP server and then use the MCP server to create a simulation.

---

Configure Environment Variables

Note: You only need to configure these variables if you want to connect to existing (external) services like Qdrant or MongoDB, or override the defaults.

All variables are set in .env.example (copy to .env and edit as needed):

Variable	Description
VITE_DASHBOARD_API_URL	Dashboard API URL for the UI (default: http://localhost:6644)
SPRING_PROFILES_ACTIVE	Spring profile for open-responses (default: platform)
OPEN_RESPONSES_STORE_VECTOR_SEARCH_QDRANT_HOST	Qdrant host (service name or external host)
OPEN_RESPONSES_STORE_VECTOR_SEARCH_QDRANT_API_KEY	Qdrant API key (if using managed Qdrant)
OPEN_RESPONSES_STORE_VECTOR_SEARCH_QDRANT_USE_TLS	Use TLS for Qdrant (true/false, default: false)
OPEN_RESPONSES_STORE_VECTOR_SEARCH_COLLECTION_NAME	Qdrant collection name (default: open_responses)
OPEN_RESPONSES_MONGODB_URI	MongoDB URI (service name or external URI)
OPEN_RESPONSES_MONGODB_DATABASE	MongoDB database name (default: open_responses)

---

Accessing the Application

• Platform UI: http://localhost:6645
• Open Responses API: http://localhost:6644

---

Cleanup

If you need to remove all containers and images for a fresh start:

# Stop and remove all containers for this project
# (use the same profiles as you used for 'up', if needed)
docker-compose down -v

# Remove any containers using the images (force)
docker ps -a --filter ancestor=masaicai/open-responses:latest --format "{{.ID}}" | xargs -r docker rm -f
docker ps -a --filter ancestor=masaicai/platform-ui:latest --format "{{.ID}}" | xargs -r docker rm -f

# Remove the images (force)
docker rmi -f masaicai/open-responses:latest masaicai/platform-ui:latest