The outage, the broken link, the 20% increase
Jason Lemkin runs SaaStr with three humans and 20+ AI agents. The company stopped hiring human SDRs and AEs because agents handle the work. So when Adobe Marketo went down for a day and a half, missed a 450,000-subscriber newsletter send, and shipped a broken unsubscribe link for two weeks, the question was not whether a human could fix it. The question was whether an agent could. The answer: no.
Marketo then came back with a 20% renewal increase. SaaStr pays $60,000 a year for the platform.
Lemkin posted the breakdown on Twitter: outage, missed send, broken compliance link, and a price rise. His take: "This is why pre-AI SaaS is dying."
What agents hit when they tried to operate Marketo
SaaStr runs marketing ops through API calls. Marketo does not support that workflow. The platform has no real webhooks, single-digit bulk exports per day, 90-day log retention, and rate caps that lock you out until tomorrow. Auth breaks on whitespace with no useful error. There is no counts API, no MCP server, no SDK, no agent toolkit.
Lemkin is not just commenting on AI disruption. He is running an org that replaced a 10-person sales team with agents. When the tooling cannot be operated by an agent, it gets replaced.
The Adobe context
Adobe Marketo is a legacy marketing automation platform inside Adobe's Experience Cloud. It competes with Salesforce Marketing Cloud, HubSpot, Oracle Eloqua, and Braze in the enterprise B2B stack. Lemkin is a former Adobe exec and co-founder of EchoSign, which Adobe acquired. SaaStr spends roughly $10,000 a year on Salesforce and a similar amount on Marketo for community ops.
The relevance for ANZ sales teams: if your org is running AI agents for GTM ops and your vendors cannot be operated by those agents, you will churn. If your vendor goes down, breaks compliance tooling, and asks for a 20% increase, you will churn faster. The question is not whether AI replaces SaaS. The question is whether your SaaS vendor can be operated by AI.
Marketo is not dying because of AI. It is dying because it cannot work with AI, and the company behind it does not appear to care.