AI and exit value · Answered by Golden Section from more than 400 B2B software companies observed
Build AI into the outcome your customer already pays you for, price it deliberately, and put a number on it. Our research across 2026 found that AI strengthens software whose value lives in proprietary data, compliance and workflow depth, and hollows out software whose value lives in its interface. In the Q2 2026 earnings season, no vertical company in our 43-name review froze, but only those that turned AI into something customers buy were re-rated. So name the outcome in your customer's own units, attach AI to it, and choose outcome, bundled or usage pricing on purpose. Instrument AI revenue, attach rate and AI-influenced retention before anyone asks. Next step: write the one-sentence outcome your AI moves, and measure it for three customers.
AI is a way to deepen the data and workflow position you already hold, not a new product category to chase. Monetize it against a named customer outcome, and keep most engineering on the core while a small team builds what could replace it.
| Metric | Value | What it means | Source |
|---|---|---|---|
| Vertical vs horizontal SaaS growth | ~32% vs ~12% | annual revenue growthFebruary 2026 market data in GS's SaaSpocalypse report; the selloff hit horizontal software hardest | Golden Section, publishedNavigating the SaaSpocalypse |
| Vertical names that froze, Q2 2026 | 0 in a 43-company review | companies reporting a budget freeze in net-new and expansionInsulation held, but AppFolio fell about 30% year to date on a beat and raise; upside came only where AI was priced | Golden Section, publishedWhere the AI Budget Lands |
| AI-validated premium | about 2–3x revenue multiple | multiple of AI-validated names over AI-exposed laggardsServiceNow near 7x revenue against Pegasystems near 2.5x, Q2 2026 | Golden Section, publishedWhere the AI Budget Lands |
| Engineering allocation | 45% core / 45% AI-native features / 10% skunkworks | share of engineering and product resourcesGS's proposed split for incumbent SaaS; the 10% builds what could cannibalize the core | Golden Section, publishedThe Enhancement Doctrine |
| AI-native retention | GRR 40%, NRR 48% | median gross and net retention, 200 AI-native companiesChartMogul 2025 panel as cited by GS, against 82% net for B2B SaaS in the same panel | External benchmarkChartMogul 2025 via Investing in Software addendum |
The threat is real and it is specific. When the interface becomes natural language, a moat built on a well-designed screen gets shallow fast, and per-seat pricing weakens wherever an agent does the work of several users. What a general model cannot copy from a cold start is years of industry data, regulatory infrastructure and workflow context, and that is where most durable vertical companies already live. This is the argument we made in The Enhancement Doctrine, and the Q2 2026 earnings season tested it: vertical software was insulated, but the market paid only for AI that had a revenue line next to it.
So the work is concrete. Start from the customer ROI you already sell and find where AI moves it. Rebalance the roadmap toward that. Choose a model with the AI pricing play, sequencing any change to the existing base, because even a correct pricing move can reset growth optics for several quarters. Then add the AI line to your KPI dashboard. Build model portability too; our September essay argues that rate limits from model providers are close to certain.
A company at $5M in annual revenue sells scheduling software to home-health agencies. A well-funded AI-native competitor launches a chat interface for the same job. The founder's first instinct is to rebuild the front end. Instead the team names the outcome customers already measure, missed visits per 1,000 scheduled, and ships an agent that uses eight years of the company's visit and compliance data to fill gaps before they happen. It launches bundled into the top tier for new logos, with a missed-visit dashboard, and moves to renewal upsells after two quarters of evidence. The board sees attach rate, AI gross margin and retention in upgraded accounts. All figures are invented for illustration.
If your product's value really is the interface over commodity data, the doctrine cuts against you and the honest response is a harder rethink of what you own. And very small companies may be right to absorb inference cost inside existing tiers for a year rather than build pricing they cannot yet support.
From the Golden Section mistakes list, each paired with the play that prevents it.
AI-native entrants and model providers change the ecosystem quickly, and founders who are not watching it get repositioned by someone who is.
Bolting a token meter or a new tier onto an AI feature without customer value behind it is hip-shot pricing at higher stakes.
AI magnifies the people using it; products that promise to remove the human entirely disappoint and churn.
Selling an AI roadmap before it works is selling phantom product, and customers remember at renewal.
In the order we would run them. Each is on its own page, most with a free Excel template.
Names the monetized outcome AI has to move before anything is built or priced.
Rebalances the roadmap between protecting the core, AI features in existing workflows and a small skunkworks.
Chooses outcome, bundled or usage pricing and sequences the change so the existing base is not spooked.
Puts the AI line on the dashboard so there is a number to disclose when a board or buyer asks.
Packages domain mastery as services, the moat that holds when software gets cheap to build.
The Enhancement Doctrine Our full argument on why AI strengthens software with a domain moat and what incumbents should do about it, with the earnings evidence linked from it.
Seat-based revenue weakens where agents replace users, pricing shifts toward outcomes and usage, and a new cost line appears in gross margin. Our Q2 2026 review found the market rewarding companies where AI adds to the number and punishing those whose customers' budgets it raids, regardless of growth rate.
It is killing weak SaaS, meaning horizontal tools with shallow workflow and interface-deep value. Software with proprietary data and regulatory embedding has so far been insulated; no vertical company in our 43-name Q2 2026 review froze. The honest caveat is that insulation has been priced and upside mostly has not.
Yes, when each one moves an outcome the customer already values and you can measure it. A feature that cannot be tied to that outcome belongs in a tier bundle or not at all, and one you cannot yet deliver should not be sold.
Compete on what they lack: years of domain data, compliance infrastructure, integrations and delivery expertise. AI-native companies grow fast but, in the panel data we cite, retain far worse, so embed AI in the workflow customers cannot pause and let retention do the defending.
Rebuilding a product around AI is the kind of structural change minority equity exists to fund, because it takes quarters to show up in revenue and no lender will underwrite it. We invest in vertical companies whose data and workflow position AI strengthens.
Growth equity →Reviewed by Dougal Cameron, CEO & Co-Founder on 2026-09-23. Golden Section observations are labeled separately from external benchmarks and illustrative arithmetic.