Quickstart

Get a project running locally in about three minutes — then try a memory write.

1. Start the backend

Requires Postgres 15 with the pgvector extension installed (use the pgvector/pgvector:pg15 Docker image), and Redis for caching.

git clone https://github.com/your-org/nubase
cd nubase

# Optional: set the LLM key used by the Memory module
export OPENAI_API_KEY="sk-..."

mvn spring-boot:run

2. Open the Studio

Visit http://localhost:3000, sign up as a platform user, and create your first project. Provisioning creates a dedicated Postgres database for the tenant with auth / storage / mem schemas and a service-role apikey.

3. Make a table

Open SQL Editor in Studio (or use psql) and create a business table:

create table public.todos (
  id bigserial primary key,
  text text not null,
  done bool default false
);

4. Hit the REST API

Every table gets a PostgREST-compatible endpoint automatically. Use the project's apikey from Studio Settings:

curl http://localhost:9999/rest/v1/todos \
  -H "apikey: $NUBASE_ANON_KEY"

5. Write your first memory

The Memory module turns conversation messages into durable, queryable facts. The LLM extracts facts and decides ADD / UPDATE / DELETE / NONE per fact.

curl -X POST http://localhost:9999/mem/v1/memories \
  -H "apikey: $NUBASE_SERVICE_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "userId": "user-42",
    "messages": [{
      "role": "user",
      "content": "I prefer steak over sushi, and my dog is named Mochi."
    }]
  }'

curl -X POST http://localhost:9999/mem/v1/search \
  -H "apikey: $NUBASE_SERVICE_KEY" \
  -d '{ "userId": "user-42", "query": "what do they eat?" }'

6. Manage it in Studio

Studio's Memory tab gives you a list view with stats, search, per-memory history (ADD/UPDATE/DELETE), an entity browser, and a Settings page that shows the live LLM/embedding config plus danger-zone actions (migrate, reset).

Where next

• The eight modules — how data, auth, the deploy layer (Assets · Functions · cron), AI Gateway and Memory share one auth model and one tenant DB.
• Memory deep-dive — update / history / entity inspection from curl.