Work
Blacksky Agent Platform
Project Type
Multi-Agent AI System — Cloud-Based
Overview
A cloud-based platform hosting specialized AI agents that extend Maurice's capabilities. Each agent handles a specific domain — starting with a Research Agent for B2B lead research. The platform follows the same local/cloud development pattern as Maurice and deploys to Railway.
Architecture
Agent Design
- Shared base agent logic with specialized agent implementations
- Single Anthropic Claude API connection shared across all agents for cost efficiency
- Each agent gets its own endpoint and can be called independently or by Maurice
- Modular design — new agents can be added without modifying existing ones
Research Agent (First Agent)
Purpose: B2B lead research and company analysis for Blacksky sales outreach.
Capabilities:
- Research companies and identify key decision makers
- Analyze company technology stacks and needs
- Identify potential pain points that Blacksky can solve
- Provide structured intelligence reports for sales outreach
- Assess company readiness for AI adoption
Future Agents (Planned)
- Content Agent — generate marketing content based on research findings
- Outreach Agent — draft personalized outreach messages
- Analytics Agent — analyze conversation patterns and conversion data
Technical Details
- Framework: FastAPI (Python 3.11+)
- LLM: Anthropic Claude Sonnet
- Deployment: Railway (production), local dev with hot reload
- Configuration: Environment-based (local vs. production)
- Integration: Called by Maurice for enhanced sales intelligence
Key Features
- Cost-efficient shared LLM connection
- Modular agent architecture for easy expansion
- Local development with production parity
- Structured output format for downstream consumption
Status
Architecture designed. Research Agent specification complete. Infrastructure ready for deployment on Railway.