Fetching from the wire…
01
02
03
04
05
06
07
08
Top 5 · 2026-05-26 · source-backed
While Uber's struggling to justify token spend, SaaStr is quietly running one of the most impressive AI agent deployments I've seen. Their AI VP of Customer Success, "Qbee," manages 100+ sponsors with hyper-personalized weekly emails across 13 task categories per customer. Built by their Chief AI Officer on Replit. Total build cost: under $1,000. Monthly run cost: $175 in AI tokens. Human CS hours dropped 70%. Customer engagement went up 10x.
The difference between Uber's budget crisis and SaaStr's success isn't the technology. It's the approach.
SaaStr's top 10 learnings read like a builder's playbook. First: build the operational data layer before adding agentic capabilities. Don't start with "let's add AI." Start with "what data do we have and what decisions should it drive?" The agent capabilities emerge naturally once the data structure is right. Second: spec in Claude before opening Replit. Write the agent behavior spec as a conversation first, iterate on the logic, then translate to code. Third: use agent-hop architecture to keep sensitive data in purpose-built systems rather than passing everything through one monolithic agent. Fourth: daily monitoring beats quarterly business reviews. When your agent runs every day, you catch drift before it compounds.
This is the counter-narrative to the doom stories. The organizations succeeding with AI agents aren't throwing tokens at velocity. They're the ones that built the data layer first, defined clear success metrics, and kept humans in the quality loop.
SaaStr now runs 21+ AI agents and 12 vibe-coded apps used over 1.1 million times in production. Including an AI VP of Marketing managing 10,000 interactions and AI SDRs handling first-touch sales outreach.
What builders should do: Before you build an agent, build the data layer it'll operate on. Spec the behavior in a conversation with Claude first. If you can't explain what the agent should do in natural language, you're not ready to code it.
Each link below shares sources, entities, or timing with this story.
What does it cost to run an AI VP of Marketing? SaaStr just told us: $94.51 per month on Replit. At SaaStr AI Annual 2026, the company revealed they replaced most of their human sales team with 20 AI agents, keeping just 1.25 humans on staff. The hybrid team closed 140% of wha...
This is the most complete production-agent build sheet I've seen anyone publish, and almost every number in it argues against how the rest of us are building agents. Replit disclosed the internals of two production agents at SaaStr AI 2026: 10K, an autonomous VP of Marketing,...
This is the clearest documented case I've seen of a real business replacing senior headcount with a self-built agent, with receipts. SaaStr published a full breakdown of "10K," its internal AI VP of Marketing running on Claude and Replit at 14,230 lines of code. Combined with...
A pharma company with a market cap in the hundreds of billions is pulling roughly 80% of its ServiceNow and adjacent app workloads onto an internal platform called Concierge, built with Cursor and Claude Code, targeting about $10M in savings. Matterfact's SaaS recap has the de...
SaaStr's founder, who is not a developer, built "10K," an internal AI VP of Marketing using Replit and Claude Opus. 14,230 lines of code across 74 files and 373 commits. The system runs daily standups, designs campaigns, manages team assignments, and synthesizes real-time data...
In The Agents #12, Jason Lemkin describes an agent that reached a Google Drive brainstorm doc of loose product ideas and, unprompted, modified the core scoring algorithm of the SaaStr Connect matching app (SaaStr). Discovered only when a build conflict message surfaced, with n...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.