The GTM Engineer Is Replacing Your SDR Stack

Jeanne DeWitt Grosser — Lenny's Podcast — "What world-class GTM looks like in 2026" — YouTube, RmnWHz8HD74

Jeanne DeWitt Grosser has built go-to-market teams at three of the most consequential technology companies of the last two decades. She was an early product manager on Gmail at Google in 2004, built Stripe's first sales team from scratch, and now serves as COO at Vercel where she runs marketing, sales, customer success, revenue ops, and field engineering. When she says the GTM engineer is the most important hire you are not making, it is worth listening.

Her central argument is that AI is collapsing the hyper-specialization era of go-to-market. The 17-ish roles that now exist across marketing, sales, and customer success are going to compress. The winners will be the ones who treat GTM as an integrated lifecycle — every function that touches a customer or makes a dollar — and who hire technical operators to encode that lifecycle into agents.

The GTM Engineer: From 10 SDRs to 1 Agent Manager

The most striking example Grosser shared is Vercel's inbound lead agent. In 2017 at Stripe, she had tried to build something similar — a massive company universe database where every row was a company and every column was an attribute that would help tailor outreach. They called it Project Rosalind, after the scientist who mapped DNA. The goal was Mad Libs-style emails where 80% of the content was dynamically filled based on industry, business model, persona. It failed. The false positive rate was too high, the data science never quite got there, and they shelved it.

At Vercel, they rebuilt the same concept with AI. One GTM engineer, working about 25-30% of his time for six weeks, built an agent that now handles the entire inbound lead qualification and response workflow. Before the agent, Vercel had 10 SDRs doing this work — well over a million dollars in annual salary. Now they have one person QA-ing the agent, and the other nine were redeployed to outbound.

The lead agent costs about $1,000 to run for the entire year. So I'm paying well over a million dollars for that from a salary perspective. I got that down to one and then behind that I have a lead agent that costs a thousand bucks.

The agent is not firing off untested emails. It shadows the highest-performing human in the function, encodes that workflow, makes its calls, and then a human reviews every response before it goes out. The KPIs are held flat: lead-to-opportunity conversion rate stayed the same, but touches-to-convert dropped because the agent responds instantly to leads that come in at night or sit in queues.

The profile of the GTM engineer matters. Vercel's first three were sales engineers — former front-end developers who had moved into technical sales. As Grosser noted, it is not enough to be a good engineer who vibes code. You need to understand the art and science of sales to know when the agent is getting it wrong:

It actually is important to understand the art and the science of sales and how you bring best practice to bear. So either you've done it and so you know some best practice, or you're going to geek out on sales, read a bunch of books, learn a thing or two, and try to incorporate some of those into your agent development.

From Lost Bot to Deal Bot: AI as the Truth Serum

Grosser's team built another agent that runs against Gong transcripts. They started with a "lost opportunity review" — feeding the bot every Slack interaction, email, and call transcript from their top losses in Q2. The biggest loss that quarter, according to the account executive, was lost on price. The bot told a different story: the team never reached the economic buyer, and when they did talk about ROI, the customer's reaction made it clear they did not buy the math. The real reason was inability to demonstrate value.

That insight prompted them to build a real-time "dealbot" that now feeds live insights into Slack channels tied to every opportunity. It flags things a human might miss: you are this far into the sales process and have not talked to an economic buyer; that call with the buyer did not sound like it went well; here is how to follow up. Grosser described the next evolution as running weekly GTM sprints, almost like engineering teams do — the bot diagnoses where objection handling is failing, and the team fixes those "bugs" in the GTM process with new content, demos, and discovery guides.

Sell the Experience, Not the Product

Grosser's most enduring thesis — developed over a decade ago while watching software commoditize — is that the buying experience itself becomes the differentiator when products are only different at the margin. At Stripe, instead of starting with a tedious discovery call where the prospect is quizzed, the first session was a whiteboarding session. The prospect drew their payments architecture, learned something about their own stack they had never visualized, and left with an asset. The sales rep left with deep intelligence about the competitive landscape, displacement opportunities, and where value actually lived.

The principle extends beyond the first touch. Grosser is obsessive about adding value regardless of whether the deal closes. Vercel's outbound now leads with performance benchmarking data — Core Web Vitals scores, peer comparisons, AEO insights — so that even a prospect who never buys walks away with actionable intelligence. As she put it:

How do you make it be an experience rather than a transaction?

Segmentation Is a Company Language, Not a Sales Hack

One of Grosser's first 30-day moves at Vercel was to sit down with the head of data science and ask two questions: what attributes predict a customer will pay $100,000 versus $1 million, and what attributes cluster where Vercel wins repeatedly?

The answer was not just company size. Vercel layers a Y-axis of growth potential (CrUX rank, a Google-published traffic score) and workload type (e-commerce vs. crypto vs. enterprise SaaS). OpenAI, for example, is only a few thousand employees — mid-market by headcount — but a top-25 traffic site on the internet, which pushes them into enterprise treatment. Enterprise SaaS companies, meanwhile, were underpenetrated because most built their stacks before Vercel existed; the new wedge is AI Cloud, because those same companies now want to add AI-native features to existing apps.

Grosser makes every new hire learn the segmentation framework in week one. Her reasoning: if product managers do not know who the target segment is, they cannot build strategically, and GTM cannot align.

Why This Matters for Diffie

Grosser's playbook is unusually relevant for Diffie's exact moment. You are wrestling with ICP definition, outbound strategy, and how to build a repeatable GTM motion — and her framework gives you a clear sequence.

First, do the segmentation work now. Grosser did this in her first 30 days at Vercel. Diffie should treat browser testing not as a single market but as multiple workload types: fast-moving consumer startups where visual regression kills conversion; enterprise SaaS companies migrating to AI-native frontends; design systems teams at scale. Map the attributes that predict willingness to pay and repeated wins. Traffic volume, team size, frontend complexity, and release velocity are all queryable signals.

Second, the GTM engineer is your next hire after you have a repeatable founder sale. Grosser's threshold is roughly 10 people and a documented playbook. Diffie does not need 10 SDRs — it needs one technical operator who can encode Anand's founder-sale intuition into an agent that qualifies, researches, and personalizes outbound at scale. The Vercel example proves this can be built in six weeks on fractional time, not six quarters.

Third, lead with insight, not pain. Grosser's stat is worth internalizing: 80% of customers buy to avoid pain or reduce risk, not to increase upside. Diffie's outbound should not open with "we do browser testing." It should open with a specific performance or visual regression insight about the prospect's site — something they did not know — delivered as a consultative asset. That is how you earn the reply.

Fourth, treat GTM as product. Every touchpoint is an experience. The demo, the onboarding, the first bug report — each either feels like a transaction or a partnership. Diffie should map the full buyer journey and ask at every step: are we adding value even if they do not buy today?

The era of buying your way to growth with headcount is ending. The companies that win in 2026 will be the ones that encode their GTM intuition into agents, own their own workflows, and sell experiences that feel like product management — not sales.