SaaStr's Salesforce Data Grew From 5GB to 40GB in 30 Days — Almost None of It Typed by a Human

SaaStr’s Salesforce storage grew from 5GB to roughly 40GB — about 21 million records — in about 30 days, almost entirely from AI agents writing data via API rather than humans typing it in. Jason Lemkin, drawing on the 20-plus production agents SaaStr runs across its own eight-figure business, says that volume is trivial on Postgres (roughly 1,000x cheaper) but expensive under Salesforce’s legacy per-record pricing — a mismatch he expects more RevOps teams to hit as agent-driven data entry scales. On the build side, a SaaStr team member built a custom renewal agent in half a day, pulling from Salesforce, WordPress, social accounts, podcasts, and Bizzabo, that produced 20-30 custom sponsor renewal decks automatically versus the five the team could manage manually — and got stronger response rates from silver-tier sponsors than diamond-tier ones. The agent still occasionally hallucinated numbers despite explicit guardrails, and Lemkin flags a real financial risk: vendors raising per-call API prices during 100x data growth can turn modest rate hikes into 200-2,000% cost increases.

The system-of-record question isn’t abstract — it’s already displaced a real partnership. ServiceTitan terminated a nine-year integration with Podium affecting roughly 1,000 shared customers, after Podium’s AI agent business grew from zero to $100M ARR in under two years and it began marketing its own field-service-management product with a 14-day migration path, competing directly with ServiceTitan’s core offering. ServiceTitan had already published marketplace rules in June barring partners from using the integration to displace its product, and updated API terms restricting which endpoints agents could call independently — guardrails written specifically because an agent-driven partner had crossed from complement to competitor.

Both stories point at the same shift: once agents write directly into a system of record at volume, the platform’s pricing model, API terms, and partner boundaries all get tested at once — tooling choices RevOps and GTM leaders are making now, not decisions to defer until the volume arrives.