The Numbers
Vercel launched a lead qualification agent in August 2025. Six weeks later, the company reduced its SDR team from 10 people down to one person running the US market, plus 20% of a person covering Europe and APAC combined.
The agent costs $5,000 a year to run. Infrastructure and tokens. It takes 20% of one engineer to maintain. COO Jeanne DeWitt Grosser puts the ROI at 32x: you saved 10 salaries, replaced them with $5,000 of compute, and the thing runs 24/7 with faster speed-to-lead.
SDR quotas went up 30% that quarter. The 10 people moved into higher-value roles. Worth noting: this was not a demo. This is production, at scale, handling real lead volume.
How They Built It
A GTM engineer shadowed Vercel's best SDR for days. Every tab she opened: LinkedIn, BuiltWith, the company site, the CRM, Slack history. The engineer turned each step into a tool-calling workflow. They documented the entire process before any AI touched it.
The agent ran in shadow mode for six weeks. The best SDR reviewed every output and fed corrections back. The loop ran until she could not improve it anymore. The architecture mirrors exactly what she did manually, except now it performs like a 90th-percentile rep 100% of the time.
A single engineer prototyped the first version over a weekend. Production six weeks later. They then ran the same framework across 30 different SDR workflows: event follow-up, product-qualified-account flows, time-based campaigns.
The Build Method
Every internal agent at Vercel gets built the same way. Three people:
- A GTM engineer
- A data scientist
- The single best subject-matter expert for that function
They document best practice first, then encode it into workflows that become the agent. A human stays in the loop to QA every output. The agent does not autonomously execute. Over time, as the subject-matter expert runs out of feedback, you pull the human.
Jeanne DeWitt Grosser ran go-to-market at Google and Stripe for roughly a decade each before joining Vercel as COO in June 2025. Six weeks into the job, she stood up a GTM engineering team with one mandate: bring agents to everything in go-to-market. Ten months later, the team has automated real chunks of core company functions.
The customer support agent now handles 93% of total case load. The content agent did 96% of major content updates last quarter. The lead qualification agent took a 10-person function down to 1.25 people.
What This Means for Sales Teams
Vercel is a $9.3 billion company that grew from $1M ARR in 2019 to $200M ARR by mid-2025. They hit $340M GAAP revenue run-rate as of March 2026. This is not a scrappy startup experimenting with AI. This is a scaled infrastructure company running production workloads.
The distinction Vercel makes internally: this is not "fire your team." This is "stop having your best people do the part a workflow can do better 100% of the time." The agent took the deterministic part of the job. The humans went up the stack.
That framing matters for how you sell this internally. The build was not hard. The discipline was. Document the human, then encode the human, then QA the agent until it beats the human, then remove the human.
Worth watching if you are running SDR teams or evaluating automation ROI. The $5,000 annual cost is the attention-grabbing number. The 32x ROI and 30% quota increase are the numbers that matter.