Vercel COO: How One SDR Now Does the Job of 10

The Rise of the Go-to-Market Engineer

This is Jean Grosser - former Stripe CPO who built their first sales team, now COO at Vercel - explaining what go-to-market looks like in the AI era. The numbers are striking.

"We went from 10 SDRs to 1 in 6 weeks." Vercel built an inbound lead agent with a single GTM engineer spending 25-30% of his time on it. The agent evaluates leads, does deep research, crafts responses. One SDR now QAs the agent instead of 10 SDRs doing the work manually. The other 9 moved to outbound.

Project Rosland was 8 years ahead of its time. In 2017, Jean tried to build a company universe database at Stripe - every company on Earth with attributes that enable personalized outreach. "We were trying to do Mad Libs with 80% fill-in-the-blank." The false positive rate on data science was too high. Now at Vercel, they're rebuilding the same thing and "it actually works" because AI can handle the fuzziness.

The go-to-market engineer is an emerging role. It's not just configuring Outreach or Salesforce. GTM engineers shadow the highest-performing SDR, document their workflows (7 tabs open, LinkedIn lookup, ChatGPT query, database checks), then encode that into agents. The goal: take sellers from 30-40% customer-facing time to 70%.

"If you are not prepared to pay 99 cents for us to perfectly solve your customer's problem, we need to wrap this up." Jean's philosophy: value-based pricing, not cost-based. When Vercel's agent costs dropped, margins improved. Customers don't care about your costs.

10 people is when you need a GTM playbook. You can't apply GTM engineering without a documented best practice. But this is forcing companies to be more rigorous earlier - which is ultimately healthy.

10 Lessons From Vercel's AI Sales Transformation

• 10 SDRs → 1 in 6 weeks - Agent handles inbound qualification; one SDR QAs
• GTM engineer - Shadows top performers, encodes workflows into agents
• Project Rosland - 2017 company universe at Stripe; works now with AI
• Lead-to-opportunity rate held flat - Agent as good as humans, faster touches
• Sellers at 30-40% customer time - Goal: get to 70% with agent support
• "Mad Libs" outreach - 80% fill-in-the-blank from company attributes
• Enterprise prospecting still hard - Multiple layers, triangulation; last to go
• Value > cost pricing - 99 cents per resolved issue regardless of model costs
• 10 people = playbook time - Can't automate without documented best practice
• Consultative selling - AI era means even more relationship-based, not less

What AI Agents Mean for Sales Team Structure

The go-to-market playbook is being rewritten. One SDR now does the work of 10 because AI handles qualification, research, and drafting. The new role - GTM engineer - shadows top performers and encodes their workflows into agents. Sales isn't being replaced; it's being automated down to the relationship.