ENTERPRISE SOFTWARE
B2B SaaS accounts for over 60% of European venture capital. The failure modes are well-known. They're still being missed.
Agentic AI is reshaping SaaS unit economics faster than most founders have updated their pitch decks. Per-seat pricing is under structural pressure. The technology that made a company fundable at Series A may be the thing that limits it at Series B. We tell you which before you commit.
Why it’s different
SaaS looks transparent. The risks hide in the architecture.
B2B SaaS is the most analysed sector in European VC — and still the most commonly over-valued at the technical layer. Multiples are tightening, agentic AI is compressing seat counts, and the market is enforcing a Rule of 40 discipline that punishes companies with hidden scaling costs. A pitch deck full of strong NRR and a Kubernetes architecture diagram does not tell you whether the product can grow without the founding engineer.
01
The pricing model may be the biggest technical risk in the company
The shift from per-seat to outcome-based pricing is not a commercial decision — it is an architectural one. Companies built on user-count logic often have that logic embedded across their billing infrastructure, API rate limiting, permission models, and usage analytics. When one AI agent can do the work of five users, seat-based revenue compresses. Companies that can re-architect around outcomes retain pricing power. Companies that cannot face a revenue cliff at exactly the moment their largest customers adopt AI tools internally. We assess whether the codebase can support the pricing model the market is about to demand.
02
Scalability claims are almost always about infrastructure, not operations
Nearly every SaaS company at Series A can show you a cloud-native architecture diagram. What the diagram does not show is the onboarding runbook that requires two engineers and four weeks of customer-side configuration. Technical scalability and operational scalability are different things. We distinguish them. The operational bottleneck is where post-investment growth stalls.
03
The founding engineer's departure is priced into the round, but not the risk
In European SaaS companies under €10M ARR, the lead engineer is typically the only person who understands the system at depth. Documentation is sparse. Test coverage of 5–25% across core services is standard. When that person transitions out of an IC role post-investment, delivery slows, bugs surface, and roadmap execution falls behind projection. We quantify the key-person exposure before it becomes a board problem.
Assessment Areas
Where we focus in B2B SaaS engagements.
AI in B2B SaaS
Every SaaS pitch has an AI chapter. Most misread the risk.
Q1 2026 saw what analysts labelled a SaaSpocalypse — a 15–20% correction in B2B software valuations driven by seat compression as enterprise customers deployed AI agents internally. The companies that held value were those whose moat was in data, workflow integration depth, or proprietary vertical logic — not in user interface.
Opportunities we verify
Proprietary workflow data as a moat. A SaaS platform that has accumulated years of customer workflow data holds something a general-purpose LLM cannot replicate. We assess whether that data is structured, clean, and legally owned — and whether the product roadmap is actually built around it.
AI as a distribution advantage. SaaS companies that embed their functionality into AI agents — rather than simply replacing their UI with a chatbot — are positioned to grow usage even as seat counts decline. API-first, well-documented companies can plug into emerging AI orchestration layers. Monolithic products cannot.
Vertical AI that incumbents cannot build. Horizontal AI platforms are becoming commoditised. Vertical SaaS with deep domain logic — regulatory workflows, compliance-heavy processes, sector-specific data models — can build AI features that generic tools cannot replicate without years of domain training.
Risks we surface
LLM-wrapped products with no defensible layer underneath. Companies that have built their AI feature as a thin wrapper over a public LLM API — no proprietary training, no fine-tuning, no structured data layer — have a moat that erodes the moment a competitor calls the same API.
Seat compression is real and underpriced in most models. The standard SaaS growth model assumes seat expansion follows user growth. When AI agents replace users for specific workflow tasks, seat expansion reverses. We model this against the current customer base and evaluate whether NRR projections reflect actual AI adoption trajectories.
Technical debt that makes AI integration impossible on the stated timeline. Companies that want to add AI-native features on top of a monolithic, undocumented codebase face a re-architecture cost that is rarely visible in standard DD. The roadmap says 'AI-powered by Q3.' The engineering reality is 18 months of refactoring first.
Know what you’re backing before you commit.
X-Ray delivers a full product and tech verdict on any B2B SaaS target in one business day — 250+ automated checks, signed off by operators who have built and scaled software companies.
250+ European engagements · 100% partner repeat rate