Thesis
Criteria
Approach
The Founder Experience
Companies
Team
News
Pitch Us
Thesis
Criteria
Approach
Companies
Team
News
Pitch Us
Back to thesis

Services-to-Software

Turning a people-powered service into software margins.

For every dollar spent on software, six are spent on services. AI unlocks that $5T+ market at software margins, and it's now a full thesis category.

Services-to-software companies look like services on the outside but run AI economics inside: 75%+ gross margins on what used to take armies of billable humans. They compress the services pyramid — outcome-based pricing replaces hourly billing, AI handles the volume, humans handle the exceptions. The result is software-like margins on services-sized TAMs.

The critical distinction: dispatchers versus builders. A dispatcher has borrowed margin — arbitraging the gap between what AI costs and what humans cost. When inference gets cheaper (it will), that margin compresses. A builder accumulates durable assets with every engagement: proprietary data, process knowledge, customer relationships, and traces from full task lifecycles including failures and corrections. Those traces are uniquely rich, because the company owns the entire lifecycle, not one step.

Seed investors trained on pure SaaS pattern-match past these deals — the Year 1 gross margins look wrong. But the companies that get this right address a TAM 10x larger than anything SaaS could touch, with a moat that deepens every engagement.

THE GATES WE APPLY AT IC

The core question: is the company accumulating durable assets with every engagement, or just dispatching cheaper labor?

This kills most deals — if the model is primarily cost arbitrage, the margin belongs to the cost curve, not the company. Beyond that: What does Year 1 vs Year 3 look like? Year 1 margins might be sub-50% as humans handle the hard cases; by Year 3 the loop should push past 75% as AI absorbs more complex work. Can we underwrite it without SaaS pattern-matching? And is the TAM genuinely services spend software never could touch — work that was unbiddable before because it required too much judgment? If it's just existing software spend at a lower price, that's a pricing play, not a category.

In the portfolio

Juno Tax

Juno Tax

Tax prep has always been a labor business — hours of manual data entry, economics set by billable hours. Juno automates ~90% of data entry across 90+ document types, cutting prep from 2–3 hours to 7–10 minutes, with per-return pricing replacing the hourly model. Founder David Haase is a CPA who ran his own firm and used early Juno prototypes inside it — a builder, not a dispatcher, accumulating process knowledge and traces from complete return lifecycles, including the corrections and overrides generic AI tools never see. Software economics on a massive services TAM.

Founders

David Haase, Sunny Shah, Jack Flitcroft

Location

San Diego, CA

Industry

FinTech & Crypto

Partnered

2026

Bonfire Lead

Jim Andelman

Stage

Early Stage

Founder fit

You may be a fit if your competition is people, not software, and AI lets you deliver the same result faster and cheaper — while building something durable underneath.