AI Defensibility

Know how safe a software business is from AI disruption

One score from 1 to 100, a comparison against 10 peers, and an action plan. The first version is built outside-in: no data room, no management time, no code access.

AI Defensibility transformation plan for 20 companies: current percentiles, modeled internal improvements and modeled M&A improvements, with the top 20% of private equity exits highlighted. October 2026.
AI Defensibility transformation plan: current scores and modeled improvements through internal work and M&A. October 2026.
View full-size chart

Built on experience

Sema has analyzed tech companies for investors for 9 years

2,700

companies analyzed

$2.7T

in enterprise value

16 of 20

largest private equity firms are clients

600+

companies scored for AI Defensibility

Who uses it

One question, asked first in every deal

Bankers and buyers now ask whether a software business can thrive as AI rebuilds the market. This report answers with evidence instead of opinion.

Buy-side deal teams

Screen a target before the data room opens. Know which metric to probe in management meetings.

Sell-side advisors and owners

Find the gaps a buyer will find, months before the process starts. Fix them or explain them.

Chief Technology and Product Officers

See which roadmap items a frontier lab could ship before you finish them, and what to build instead.

What you get

Score, peers, evidence, plan

1 to 100

One score built from 5 metrics and 29 sub-metrics.

10 peers

A peer board of 10 comparables, benchmarked against 249 private equity exits and 195 public software companies. The median public software company scores 55.

Every score sourced

An evidence list behind each sub-metric, plus the evidence that would raise it.

Action plan

Each gap becomes a workstream, with the modeled score lift from doing it.

The method

Five metrics

Product Defensibility

How hard is the product to rebuild with AI in 12 to 24 months?

Revenue Resilience

How exposed is revenue to AI-driven price cuts or substitution?

Business Model Resilience

Can the business model absorb an AI-native competitor?

Market Position Strength

Do market share and network position hold when AI-native entrants arrive?

Data Asset Value

Does the company own data an AI model cannot get elsewhere?

How it works

Three steps

Step 1 · Preliminary

Outside-in score. Public filings, product, patents, hiring, pricing and reviews. 0 hours of company time. 2 business days.

Step 2 · Live walkthrough

30 minutes, walked through live. Sema takes you through the score, the peer board and the action plan.

Step 3 · Adjusted

Management answers. About 20 discussion questions, about 1 hour, by call or in writing. Included in the price. Code access optional.

Example · anonymized

Medical software company: 44 outside-in, 55 with management answers

A growth-equity-backed medical software company scored 44 of 100 outside-in. Management answered the discussion questions, and the score rose to 55: the 90th percentile of 249 private equity software exits, where the median is 43.

The action plan models 57 (92nd percentile) with the plans the company already has, and 65 (99th percentile) with full execution. Modeled scores are upper bounds, not observed results.
‍

Step 1 · 44

Outside-in. Public information, 0 hours of company time.

Step 2 · 55

90th percentile. Management answers added. Measured.

Step 3 · 57

92nd percentile. The company's known plans. Modeled.

Step 4 · 65

99th percentile. Full execution of the action plan. Modeled.

Get started

See where a company stands

Name one company. We will score it and walk you through the result live.

Sema works with some of the world’s most amazing organizations