Growth Intelligence Assistant

#Overview

Growth Intelligence Assistant gives product teams, founders, and growth strategists instant structured market intelligence. Ask a natural-language question and specialist AI agents fan out across the web — gathering signals, scoring confidence, synthesising strategy, and optionally generating outreach campaigns with a closed feedback loop.

#Highlights

• 🏆 Team TerraBytes — Veracity Gen AI Hackathon (HatchWorks)
• 🤖 9 coordinated agents — 6 research + 3 execution sub-agents
• ⚡ Real-time streaming — watch agents complete live via Server-Sent Events
• 🔄 Research → Action loop — execute, collect feedback, refine with outcomes
• 🗺️ Strategic mind maps — visual synthesis of positioning, pricing, and threats

#Two-Stage Architecture

Stage 1 — Research (parallel)

• 📈 Market Trends — industry signals, category growth, emerging tech
• ⚔️ Competitive — competitor positioning, feature comparisons, SWOT
• 🔄 Win / Loss — customer sentiment, switching triggers, review signals
• 💰 Pricing — tiers, packaging strategies, value anchors
• 🎯 Positioning — messaging analysis, brand differentiation, GTM
• 🌐 Adjacent — disruption threats, adjacent markets, tech substitution

Stage 2 — Execution Engine

• ✍️ Content Agent — campaign briefs, copy angles, pain-point mapping
• 🧪 A/B Variant Agent — variants with falsifiable hypotheses tied to research
• 📬 Outreach Formatter — email/LinkedIn sequences + deployment timelines

#Features

• 💬 Natural-language growth queries with multimodal attachments (screenshots, charts, pricing tables)
• 📊 Confidence-scored findings with source attribution and signal counts
• 🔥 Adjacent-market threat heatmaps and risk timelines
• 🧵 Threaded follow-ups with full conversation context
• 💾 Persistent company memory and recoverable session history
• ⏱️ Cost & latency tracking per query (wall-clock time, API calls)
• 👍 Feedback loop — rate recommendations, record performance, refine the next cycle

#Tech Stack

#Architecture

growth-intelligence/
├── app/              # Next.js routes + API (chat, refine)
├── components/       # Dashboard, artifacts, mind map UI
├── lib/agents/       # Research + execution agent orchestration
├── lib/tools/        # Live signal / web research tools
└── supabase/         # Persistence & session memory

#Orchestrator Flow

• 1. Classify query (LLM + deterministic intent detection)
• 2. Fan out 6 research agents in parallel
• 3. Optionally fan out 3 execution sub-agents grounded in research
• 4. Synthesise findings into prose + recommendations
• 5. Generate a strategic mind map
• 6. Stream structured JSON chunks to the frontend