Agent-First
M&A

Boring Business Arbitrage

Michele Mauri@mike
01

The Problem

VC burnout is real. Chasing the next OpenAI means competing in a winner-take-all market where 95% of venture-backed startups return nothing. The math does not improve with more conviction.

Meanwhile, recession-proof, asset-backed local industries - HVAC, plumbing, water delivery, commercial maintenance - churn out consistent cash flow year after year. Institutional capital ignores them. The asymmetry is staggering.

95%VC startups that return zero
12-18xMultiple gap: buy vs. sell
02

The Solution

Acquire unsexy but stable businesses. Think V Express - water delivery services, HVAC contractors, plumbing operations. These are not growth rockets. They are cash machines. Recession-resistant. Asset-backed. Owner-operated at small scale, with administrative overhead eating 20-30% of revenue.

20-30%Admin overhead as % of revenue
2-3xEBITDA acquisition multiple
03

The Catalyst

Agent-First & MCP

The real unlock is not the acquisition - it is the transformation. Build MCP (Model Context Protocol) integrations so personal AI agents can autonomously handle discovery, ordering, scheduling, billing, and customer communication. No human-in-the-loop for routine operations. The agent is the front door.

An agent discovers your water delivery needs, negotiates the reorder, schedules the drop, processes payment, and logs the interaction - all without a single human touchpoint. That is the agent-first operating model.

1
Agent discovers need & negotiates reorder
2
Agent schedules delivery & dispatches
3
Agent processes payment & logs interaction
04

The Play

A four-step machine:

1
Buy these businesses using SBA loans at favorable ratesThe Canada Play: 10% down, government-backed financing
2
Inject AI automation to strip administrative marginsFrom ~25% to under 5% through agent-first operations
3
Roll them up under a single operating platformCentralized back-office, unified agent infrastructure
4
Exit at a multiple reflecting software-enabled recurring revenueNot fragmented local services - sell at 8-12x EBITDA

The arbitrage: buy at 2-3x EBITDA, sell at 8-12x. The gap is automation.

2-3xBuy multiple
8-12xSell multiple
4-6xMultiple expansion
~25%Admin margin stripped → ~5%
//

SBA Loan Arbitrage

The Canada Play: SBA 7(a) loans let you acquire businesses with as little as 10% down. Government-backed financing at rates that make the math work on day one. European bureaucracy makes this nearly impossible. Geographic regulatory arbitrage is itself a moat.

CA10% down, government-backed, favorable rates
EUNear-impossible bureaucracy, limited SBA-equivalent

Boring businesses. Agent-first operations. Geographic arbitrage. The stack is transparent. The math is straightforward. The execution is everything.