Investors Debate: SaaS Risk Is Not Extinction but Missing the Agent Layer

After BRZE and TTAN earnings stoked market fears of an AI shock to SaaS, investor Matt Slotnick laid out his thesis across a series of posts: the past three months of price swings don't matter — the existential challenge to application software is unchanged. A new agent layer is rising and almost no one has a good answer. The risk is not extinction but missing the agent layer, where all the growth is; fall behind a few product cycles in a fast-moving market and you may never recover.

Confirmed

• Slotnick grants the traditional logic of application-layer SORs: CRM holds customer state, HRIS holds employee state, and owning the core record confers architectural and pricing power — the standard argument that incumbent SORs win the AI era.
• But he argues this has not played out in practice: enterprises rely heavily on third-party apps for key functions, first-party add-ons are often poor, and incumbents have shipped almost no successful agent-layer products, resorting to partnerships or acquisitions.
• SORs store "flattened" state, missing the process data — human actions, emails, calls, external data — that produced it; agent activity happens outside the platform, so the SOR sees state change from x to y without knowing why or how.
• Agent-related spend therefore flows to a new infrastructure layer outside existing systems (as agent adoption has spawned new data stores like turbopuffer), where incumbents have no product and no credible path in.
• In the Salesforce/CRM-Anthropic debate, Slotnick assessed the strategic risk: he understands the motivation, but it amounts to ceding the hottest agent layer and handing workflow telemetry to the largest wallet-share competitor; current CRM data layers are ill-suited for agent applications, and Salesforce needs a better answer than letting Anthropic own the agent layer, such as rebuilding a virtualized data layer.
• Jared Kubin echoed the frame — the structural challenge to the application layer is unresolved and moats may come from attributes like frictionlessness. walnutavevalue argued CRM has no shot at the agent layer and Anthropic can't own distribution, yet enterprises won't rip out CRM: migration is too painful, returns unclear, and careers are at risk; application software lives or dies in the agent layer, though he personally bets on distribution.

Why it matters

• Slotnick frames the risk as "missing out, not extinction": systems disintermediated by agent work risk sliding into irrelevance with no right product.
• If SORs don't evolve into agent-native platforms, the agent ecosystem above them will define the requirements, spawning a new intermediary platform player — a transfer of power to a new layer. He also asks a more fundamental question: as work moves to agents, are today's records even the right records to keep?
• The debate around the CRM-Anthropic deal shows the thesis has spread from individual opinion to an industry-wide argument over the trade-off between integrating frontier models and holding the agent layer.

2026-09-09 ~ 2026-09-10 · 24 related posts

• Matt Slotnick: nothing in 3 months invalidated the agent-layer threat to app software — matt_slotnick · 2026-09-09
• Investor: the existential threat to app-layer software is the emerging agent layer, not stock swings — matt_slotnick · 2026-09-09
• [source] App software's risk isn't extinction — it's missing the agent layer growth market — matt_slotnick · 2026-09-10
• Why app-layer SORs are valuable: they own core business state — matt_slotnick · 2026-09-10
• Owning the core record confers architectural and pricing power — matt_slotnick · 2026-09-10
• The 'whoever owns the record wins AI' argument for SOR apps, stated — matt_slotnick · 2026-09-10
• SOR incumbents' AI win is theoretical so far, not practice — matt_slotnick · 2026-09-10
• Incumbent SORs have shipped almost no successful agent-layer products — matt_slotnick · 2026-09-10
• As work shifts to agents, are today's records even the right records? — matt_slotnick · 2026-09-10
• Flattened state in SORs strips out all the data that created it — matt_slotnick · 2026-09-10
• System of Records aren't built for agent-native patterns yet, argues ex-employee — matt_slotnick · 2026-09-10
• SORs see state changes but lose all context as agents work off-platform — matt_slotnick · 2026-09-10
• If SORs don't adapt to agents, a new platform layer will be built on top of them — matt_slotnick · 2026-09-10
• Agents will shift power to new layers above existing enterprise systems — matt_slotnick · 2026-09-10
• Agent infrastructure is where spend happens — and incumbents have no path in — matt_slotnick · 2026-09-10
• AI agents are building a new infra layer that leaves existing enterprise systems behind — matt_slotnick · 2026-09-10
• Investor: falling behind on agent product cycles may be impossible to recover from — matt_slotnick · 2026-09-10
• Application software's real risk is missing the agent layer, not extinction — matt_slotnick · 2026-09-10
• VCs debate the CRM-Anthropic deal: the agent layer, not distribution, decides app software's fate — walnutavevalue · 2026-09-10
• [source] CRM is a bad data store for agent apps, argues analyst on Anthropic deal — matt_slotnick · 2026-09-10
• Why letting Anthropic own the agent layer could hollow out CRM vendors — matt_slotnick · 2026-09-10
• [source] CRM Has No Chance at the Agent Layer, But Enterprises Won't Rip It Out — walnutavevalue · 2026-09-10
• Investor Jared Kubin: Agent Layer Still Poses Existential Challenge to App-Layer Software — JaredKubin · 2026-09-10

1 near-duplicate retellings: matt_slotnick