SaaStr AI went from 30 humans to 3 humans plus 20+ AI agents in production. Revenue climbed 47% year over year after being down 19% the prior period. The agents closed over $1M in deals, run win-back campaigns at 72% open rates, and send 15,000+ outbound messages monthly through Artisan. Qualified drives seven figures in pipeline.
Founder Jason Lemkin says the agents still need proper B2B software to operate. The take that AI agents can just work off a Postgres database without CRM is wrong for most companies.
First problem: humans will not log leads in Postgres. Your AE is not writing SQL. Your VP Sales is not querying tables to forecast. Your CFO is not running joins for pipeline coverage. Humans need interfaces that match how they think: pipelines, stages, territories, quotas. Strip that away and half your team cannot do their jobs.
Second: agents need shared systems more than humans do. Put 20 agents writing to a raw Postgres table and you get 20 different interpretations of what a qualified lead is. No shared forecasting rules. No consistent ownership logic. No audit trail anyone trusts. SaaStr's agents work on top of the same system of record because they need the same shared logic humans do.
Third: every other system in your stack needs a source of truth. Marketing automation, billing, BI, CS platforms, comp tools, forecasting, exec dashboards all trust your CRM. Replace that with a database that 20 agents are writing to in 20 different ways and nothing downstream works.
Fourth: Salesforce is safer. Agents make mistakes. They hallucinate deal amounts, loop on bad prompts, misclassify leads. When that happens inside Salesforce, guardrails catch it. Permissions limit damage. Validation rules reject bad writes. Audit trails show exactly what happened. When it happens in a raw Postgres DB you wired up yourself, you find out Monday morning when the forecast is broken.
Add SOX compliance, GDPR, FedRAMP, HIPAA, SOC 2, data residency, encryption, backup and recovery. You can build all that yourself or use a platform where it already exists and is already approved by every enterprise procurement team.
The winning architecture is not replace CRM with Postgres. It is expose the CRM as a platform humans and agents work on together. At SaaStr, agents outnumber humans 7:1. They still run on Salesforce. Not because they have to, but because the agents need the shared system of record even more than the humans do.
SaaStr reported $4.5M ARR in 2025, operates with single-digit headcount due to AI agent adoption, and runs a $200M AI-focused venture fund. The company has no ANZ presence, operating primarily from San Francisco.