One business model, one shared threat
The cluster's common denominator is the per-seat subscription — you pay for each human who logs in. That model built Salesforce (#1 CRM), ServiceNow (~97% subscription at a 31% operating margin), and Workday (cloud HCM, 97% gross retention), and it underlies the collaboration challengers Figma, Canva and Notion. Read their weighings side by side and the same sentence recurs in every one: agentic AI could either grow the seat (smarter software, higher price) or erase it (an AI agent does the work a seat used to). No other software cluster is this uniformly exposed to a single binary — the technology that is supposed to expand them is the same one that could hollow them out.
The incumbents are repricing themselves on the bet
The three incumbents have each made the AI question explicit and reorganized around it. Salesforce is repackaging itself around Agentforce — selling AI agents as "digital labor" priced per conversation rather than per seat, a deliberate hedge against its own model. ServiceNow is betting its workflow data and governance layer make it the place agents run, not the thing they replace. Workday's own study lands the seat-compression question as genuinely contested. The tell across all three is that growth has already cooled to low-to-mid teens (Workday +13%, Salesforce ~6-7% organic) — so the multiple now rests entirely on whether the AI repricing adds a second growth engine before the seat base stops expanding.
The collaboration challengers face a sharper version
For Figma, Canva and Notion, the threat is more existential because their seat is a canvas — and AI can route around a canvas entirely. Figma's question is whether AI disintermediates the design surface itself; Canva's is whether generative tools commoditize design and let a bigger-distribution rival (Adobe, Microsoft, Google) bundle it away; Notion's is whether "the structured workspace every agent reads from" is a durable category or a transient one before Microsoft's Copilot-plus-Loop bundle absorbs it. They have the strongest growth in the cluster (Figma +41%, 139% net retention; Canva ~$4B ARR; Notion +50%) — which is exactly why the bear case is distribution and bundling, not demand.
Toast is the one that escaped the seat
Toast is the cluster's structural outlier and its useful contrast: a vertical restaurant operating system that bills mostly on payments, not seats — ~82% of revenue is pass-through payment volume at a thin ~48-58bps take rate, with software bundled on top. It just turned its first GAAP profit at 164,000 locations and a 34% adjusted-EBITDA margin on the software layer. Placed beside the seat-based six, Toast shows the alternative monetization the others are edging toward: tie revenue to the customer's activity (transactions) rather than to a headcount an AI agent can shrink. The catch is that a thin payments take rate is competible and its SMB base is cyclical — it traded one model's risk for another's.
Where they agree — and where they split
All seven agree the underlying businesses are healthy today — high gross margins, strong net retention (119-150% at several), sticky systems of record. The disagreement is entirely about the forward: whether agentic AI is a TAM expansion they monetize or a seat contraction they can't escape, and whether today's multiples already price the optimistic answer. The incumbents bet their data and governance make them where agents run; the challengers bet their network and workflow lock-in survive bundling; Toast bets payments sidestep the question. The demand for enterprise software isn't in doubt. Whether the unit it's sold by — the seat — survives the thing being sold is the whole cluster's open question.