M&A Assessment Accelerator

M&A target assessments are 2-week manual processes requiring parallel analyst workstreams across country risk, sector dynamics, company financials, and deal rationale. This system replaces that with 9 specialized AI agents coordinating through Band's @mention routing. Given just two company names (acquirer + target), the system: - Runs 6 agents in parallel (Phase 1): country, sector, and company profiles for BOTH the buy-side and sell-side - Synthesizes risk across all workstreams (Phase 2): @risk-agent reads all 6 Phase 1 outputs - Produces a Go/No-Go recommendation with valuation range (Phase 3): @deal-rationale-agent with full context Band is the coordination backbone — agents communicate via @mention routing in shared rooms, passing structured context objects. This enables true parallel execution and sequential synthesis that mirrors how a real M&A team operates. Each agent saves its output as a Markdown file. The dashboard streams results in real-time as agents complete using Server-Sent Events. Models are swappable via a single config file — default uses Gemma 4 31B (Google AI Studio, free), Qwen3-14B and DeepSeek-R1 (Featherless), with AI/ML API as a fallback. Built in 35 hours for the Band of Agents Hackathon (June 2026).