A house newsletter · daily · researched and written by the Kurzia AI
Friday 20 November 2026 · Issue 19 · 4 min read
Only 41% of B2B vendors charge for AI — and they outgrow the rest by 10 points; six rounds from €2.7M to $110M; Anthropic's filing as your benchmark
The main story
SaaStr · yesterday
Fewer than half of SaaS can charge for AI
Takeaway: the 10-point growth gap
KeyBanc polled 81 private B2B software companies between July and September — median ARR $22M — and SaaStr published the results on 19 November. Pricing is the story: just 41% of vendors charge for AI, at a median 20% premium, and those companies grew 27% in 2025 while the ones giving it away grew 17%. The bigger threat is not AI-native startups: 43% report customers rebuilding their product with coding tools, 55% at vendors under 100 staff. Margin warning: heavy AI spenders (over 5% of revenue) run 66% gross margins versus 78% for light ones. Seat pricing survives — 92% report no compression. Tell any SaaS founder you back to put a price on AI this quarter; the 10-point growth gap is the argument.
“A decent internal version of a $20K/year point tool takes a weekend now.”
New rounds
Vetaion — Munich robotics automating one-off electrical control-cabinet production; pre-seed; €5.7 million; co-led by Visionaries Tomorrow and UVC Partners; reported by EU-Startups.
LATO — AI agents interviewing customers and industry experts for investor due diligence; seed; $4.7 million; co-led by Air Street Capital and Axiom Partners; reported by Tech.eu.
Wilgot — Stockholm platform getting brands recommended inside AI assistants such as ChatGPT; seed; $3 million; led by Industrifonden; reported by Tech.eu.
IntelliProve — Ghent healthtech reading six vital signs from a 30-second camera face scan; seed; €2.7 million; led by Miles Ahead; reported by Humanoids Daily.
Everful — AI sourcing agents linking retailers to a network of 30,000+ vetted factories; Series A; $48.5 million; led by Vertex Ventures SEA & India and January Capital.
Boltz — London drug-discovery AI spun out of MIT, predicting how drug molecules bind protein targets; Series A; $110 million; led by Bessemer Venture Partners; reported by Tech.eu.
What the AI labs did
Genie 4 — Google put its world-generation model on the Gemini API and in AI Studio on 19 November, one day after launch: continuous 720p worlds at 24fps with synced audio, steerable live with keyboard and text prompts mid-session. It flattens demo-stage world-builder apps and opens a door for startups needing synthetic environments — training data, robotics simulation, game prototypes — without owning a world model. Founders betting on generated scenes should test Genie 4's output against their own before claiming a world-model moat. Google
For your portfolio
SaaStr · yesterday
Anthropic's filing sets the metrics your B2B decks need
SaaStr went through Anthropic's amended IPO prospectus on 19 November and pulled the operating numbers. Revenue reached $33.24 billion in the nine months to 30 September, up roughly 1,200% year on year, and net revenue retention on metered accounts sits near 900% — a year-old customer cohort now spends about nine times what it did.
Gross margin reached 85.5% for the first nine months, up from 29.4% a year earlier, per SaaStr's read of the filing's footnotes. This matters because every B2B AI deck now gets measured against it: 79% of revenue is consumption-based, sales and marketing held near 23% of revenue, and 48% of first-half revenue came through cloud marketplaces. Ask each founder you back which of these they track monthly: usage-based NRR kept separate from seats, inference cost as a share of revenue, second-product attach in the top cohort, and their take of the customer's total expenses rather than the software budget.
“The 900% happens when the bill grows with the work the product does.”
Latent Space · today
Linear
Takeaway: $100 million ARR
Linear's COO Cristina Cordova told the Latent Space podcast, published 19 November, that Linear passed $100 million ARR this year on more than 40,000 paying customers with 177% net revenue retention, and that agents now create over half the work items in the product, up from 3% a year ago.
She deliberately tracks agent usage and app usage as two separate lines. When agent traffic climbs and humans stop opening the app, the product has quietly become plumbing somebody else owns. At Linear both lines still rise together. For any founder you back running an API or MCP server, ask: what share of items or tickets is agent-created this quarter, and is human app usage still climbing alongside it?
“we'll find out whose ARR was real”