Files · Databricks Agent Mesh for Small Business Workflow Automation

# Databricks Agent Mesh for Small Business Workflow Automation
 
> A multi-agent routing mesh powered by Databricks models that coordinates specialized agents for order processing, customer service, and analytics for small businesses.
 
A tutorialized reference solution from [reaatech.com](https://reaatech.com), demonstrating how to build production-grade AI systems with the `@reaatech/*` package family.
 
## What It Builds
 
A multi-agent orchestrator for small business workflows. The system routes incoming user requests to the most appropriate specialist agent based on intent classification:
 
- **Order Agent** — handles order status inquiries, creation, cancellation, and returns
- **Support Agent** — manages ticket creation, FAQ resolution, and escalation
- **Analytics Agent** — provides sales summaries, customer trends, and business metrics
 
## Architecture
 
```
┌─────────────┐     ┌──────────────┐     ┌─────────────────┐
│  Next.js 16 │────▶│  API Route   │────▶│  Agent Mesh     │
│  App Router │     │  /api/mesh   │     │  (REAA Router)  │
└─────────────┘     └──────────────┘     └────────┬────────┘
                                                  │
                    ┌─────────────────────────────┼─────────────────────────┐
                    │                             │                         │
              ┌─────▼──────┐              ┌───────▼──────┐        ┌────────▼─────┐
              │ Order Agent │              │ Support Agent│        │AnalyticsAgent│
              │ (Mastra)    │              │ (Mastra)     │        │ (Mastra)     │
              └─────────────┘              └──────────────┘        └──────────────┘
                    │                             │                         │
              ┌─────▼──────┐              ┌───────▼──────┐        ┌────────▼─────┐
              │  Drizzle    │              │  Drizzle     │        │  Drizzle      │
              │  + Postgres │              │  + Postgres  │        │  + Postgres   │
              └─────────────┘              └──────────────┘        └──────────────┘
```
 
The mesh uses REAA packages for routing, session management, gateway handling, and observability, backed by Databricks models via Mastra.
 
## Prerequisites
 
- Node.js 22+
- pnpm 10+
- A Databricks workspace with model serving endpoints
- PostgreSQL database
 
## Setup
 
```bash
pnpm install
```
 
Copy `.env.example` to `.env` and fill in your credentials:
 
```bash
cp .env.example .env
```
 
### Environment Variables
 
| Variable | Description |
|----------|-------------|
| `DATABRICKS_HOST` | Databricks workspace URL |
| `DATABRICKS_TOKEN` | Databricks personal access token |
| `DATABASE_URL` | PostgreSQL connection string |
| `API_KEY` | API key for mesh endpoint authentication |
| `OTEL_EXPORTER_OTLP_ENDPOINT` | OpenTelemetry collector endpoint |
| `LANGFUSE_PUBLIC_KEY` | Langfuse public key for observability |
| `LANGFUSE_SECRET_KEY` | Langfuse secret key |
| `LANGFUSE_HOST` | Langfuse host URL |
| `SESSION_TTL_MINUTES` | Session time-to-live (default: 30) |
| `SESSION_MAX_TURNS` | Max turns per session (default: 100) |
| `MCP_REQUEST_TIMEOUT_MS` | Agent dispatch timeout (default: 30000) |
| `MCP_MAX_RETRIES` | Max retries for failed dispatches (default: 3) |
| `ENABLE_CIRCUIT_BREAKER` | Enable circuit breaker pattern |
| `RATE_LIMIT_WINDOW_MS` | Rate limit window in ms (default: 900000) |
| `RATE_LIMIT_MAX_REQUESTS` | Max requests per window (default: 100) |
| `NODE_ENV` | Environment (development/production) |
 
## Running
 
```bash
pnpm dev            # start Next.js dev server
pnpm test           # run tests with coverage
pnpm typecheck      # TypeScript type checking
pnpm lint           # ESLint validation
```
 
## API Reference
 
### `POST /api/mesh`
 
Routes a user request to the appropriate agent.
 
**Request Body:**
 
```json
{
  "user_id": "usr_abc123",
  "input": "Where is my order?"
}
```
 
**Response:**
 
```json
{
  "request_id": "550e8400-e29b-41d4-a716-446655440000",
  "session_id": "660e8400-e29b-41d4-a716-446655440001",
  "agent_id": "order-agent",
  "response": "I found your order #123. It was shipped on March 15 and is currently in transit.",
  "workflow_complete": false,
  "duration_ms": 1423
}
```
 
## Project Layout
 
```
app/                  Next.js App Router pages + API routes
  api/mesh/route.ts   POST /api/mesh — main routing endpoint
  layout.tsx          Root layout with font setup
  page.tsx            Dashboard page
src/                  Application source code
  lib/                Database schema and Drizzle client
  mesh/               Agent registry and Zod schemas
  services/           Agent implementations and session management
tests/                Vitest test suite (mirrors src/)
packages/             API references for every dependency
```
 
## License
 
MIT — see [LICENSE](./LICENSE).