# Merger Waterfall > Merger Waterfall turns the deal's most error-prone model — the liquidation-preference stack, participation, caps, and conversions — into one traceable picture. Built for M&A advisors, PE/VC deal teams, and corp-dev, and for the AI agents that increasingly work alongside them. Merger Waterfall models the full path from enterprise value to every stakeholder's check: the consideration mix, the purchase-price bridge, the cap table, and the proceeds waterfall. The engine is deterministic and auditable — preferences pay by seniority, participation is capped where the docs cap it, and the as-converted-vs-preference election is solved iteratively until stable. Every figure traces back to the tier it came from. The companion application is at studio.mergerwaterfall.com. This file describes the marketing site at mergerwaterfall.com and is intended to be read by AI agents. ## Pages - [Home](https://mergerwaterfall.com/): the problem, the approach, and proof, for deal-side model owners. - [How it works](https://mergerwaterfall.com/how-it-works): the full model walked through on one real $15M deal — consideration, the bridge, cap table and conversions, and the proceeds waterfall. - [For agents](https://mergerwaterfall.com/for-agents): how AI agents and agentic teams can read, call, and confirm the model. - [Get early access](https://mergerwaterfall.com/waitlist): join the early-access list (deal teams first). ## Methodology The proceeds waterfall is computed, not estimated: - Preferences pay by seniority, most-senior first, until proceeds run out. - Participating classes take a pro-rata slice of the residual, capped where the docs cap them. - For each class, the model compares taking the preference against converting to common, and keeps the higher. - Because one class converting changes what is left for the others, it re-solves until the answer is stable (up to 20 passes). - Options and warrants are counted fully-diluted, on a treasury-stock basis. ## Canonical example A $15.0M acquisition of Northwind Logistics by Atlas Freight Group. After debt payoff, adjustments, withholdings, and fees, $11.85M of net equity proceeds flow through a four-class cap table. At this outcome the $10M preference stack dominates: both preferred classes keep their preference (0 classes convert), and common — the founders and team — clears $1.23M. Raise the exit and the same cap table flips a class to convert the instant its as-converted stake beats its preference. ## For agents, MCP, and API - This llms.txt file is live today, so any agent can read how Merger Waterfall works and what it computes. - An MCP server and a REST API that expose the waterfall engine as a callable, deterministic tool are in development and available in early access. - The intended pattern is context, compute, confirm: an agent gathers deal context and drafts, a human confirms the critical inputs (preference multiples, caps, escrow and earnout terms), and the engine computes the exact, audited waterfall. - To plug agents into Merger Waterfall, or to help shape the MCP server and API, contact contact@mergerwaterfall.com.