Interactive demo of a structured product generator powered by Solace Agent Mesh. Three specialist AI agents (Discovery, Structuring, Document) collaborate in real-time to produce validated structured note proposals from natural language client briefs.
Hosted on GitHub Pages. Open index.html or visit the deployed URL.
- Open the app in your browser
- Click the gear icon (top right) to configure your LLM endpoint
- Enter your OpenAI-compatible API endpoint, key, and model name
- Click "Save" and start generating
Any OpenAI-compatible /v1/chat/completions endpoint works:
| Provider | Example endpoint |
|---|---|
| LiteLLM Proxy | https://your-proxy.com/v1 |
| Anthropic (via LiteLLM) | Model: anthropic/claude-sonnet-4-20250514 |
| OpenAI | https://api.openai.com/v1 |
| vLLM | http://localhost:8000/v1 |
Browser -> OpenAI-compatible API (LiteLLM / vLLM / etc.)
|
Agent 1: Discovery (parameter extraction from natural language)
Agent 2: Structuring (product selection + business rule validation)
Agent 3: Document (term sheet, suitability, e-pricer XML)
All configuration is stored in browser localStorage. No server required.
- Orchestrator routing between specialist agents
- Business rule validation blocking workflow on failure
- Parallel artifact generation
- Timestamped audit trail
- Data governance and schema validation