AI Pitch Deck Research Agent

How Cassidy automates pitch deck research using AI

Step 1: Parse the pitch brief

The Workflow triggers when a new pitch brief is submitted—via form, email, or Slack message. Cassidy extracts structured entities including prospect name, category, geography, ICPs, channels, and timeline to define the research scope.

Step 2: Generate targeted queries and retrieve sources

Cassidy expands seed terms into long-tail queries with competitor and synonym variations, then searches across your approved source hierarchy—analyst reports, government databases, web analytics platforms, and your Knowledge Base of past pitch research.

Step 3: Extract and normalize market data

The Workflow pulls key tables, figures, and statistics from retrieved sources, capturing publication dates and URLs. Cassidy normalizes all values to consistent currency, inflation basis, and time periods, flagging any data older than your specified freshness threshold.

Step 4: Build TAM/SAM/SOM models

Cassidy calculates market sizing using both top-down (industry totals segmented by geo and ICP) and bottom-up (unit counts × ARPA) approaches, then reconciles the two with sensitivity analysis on penetration and pricing assumptions.

Step 5: Compile competitive intelligence

The Workflow aggregates traffic share, paid media tactics, product positioning, and messaging from each competitor, then auto-generates 2x2 matrices and battlecards with strengths, weaknesses, and counter-positioning.

Step 6: Assemble slides with citations

Cassidy drafts slide copy with structured placeholders, renders charts and visuals, and attaches footnotes to every statistic—source URL, access date, and methodology. A Human-in-the-Loop step allows your team to review before final export.

Step 7: Package deliverables

The Workflow outputs the core narrative deck, appendix with detailed assumptions, and a source archive with links and screenshots—ready for client presentation or further refinement.