SaaStr built an AI VP of Marketing because every vendor tool they demoed only generated content. They already have 5,000+ pieces of content across 13 years. The bottleneck was never copy. It was orchestration.
Their custom agent, called 10K, plans marketing activities, builds campaigns, manages Salesforce pipeline, and assigns daily tasks to the two humans running marketing. It runs 24/7, updates in real-time, and crunches 5+ years of historical data plus live vendor API feeds.
Founder Jason Lemkin says 10K is already better than an average human marketer. Not better than a great VP of Marketing, but good enough to run in parallel. Compare outputs. Debate recommendations. Use both.
The honest bit: zero politics, zero sunk cost fallacy, zero pet projects. When a campaign is not working, 10K flags it on day 2, not day 60. When pipeline data shows one channel outperforming another by 4x, it does not care who has been running the underperforming channel for three years.
What it cannot do yet: manage or train their 20+ third-party AI agents directly. It can analyse outputs from tools like Artisan AI SDR and Qualified, but it cannot adjust their behaviour. Every agent has different APIs, guardrails, and training interfaces. Lemkin says someone should build an orchestration layer that sits above all AI agents and manages them from a single dashboard. They will be first customer.
SaaStr spent over $500k on AI infrastructure. In 2020, they had 20+ employees and needed a second office: 3 designers, 5 in sales, 3 in content, plus ops and events. Today: 3 full-time humans, 20+ AI agents, same revenue scale.
Their 90/10 rule: buy 90% of what you need, only build the 10% where no solution exists. They tried to buy an AI marketing orchestrator. Nothing existed. So they built 10K.
For sales leaders testing AI SDR platforms, AI sales automation, or AI tools for prospecting: the gap is not content generation. The gap is orchestration, strategy execution, and honest pipeline analysis without bias. That is what SaaStr is betting on.