GTM Engineering — Why now?

I'm building a new department at Vercel called GTM Engineering. Its primary focus is applying AI to GTM problems and dogfooding the Vercel AI Cloud to do it. We also ship projects that unblock our sales and marketing teams like the v0 docs & contact sales form improvements.

We've had great early success partnering with our SDR team and are open sourcing some of our work today (Lead Agent template).

I'd like to spend some time unpacking why I think it's the right time for you to stand up a GTM Eng org at your company and how you can go about it.

Why now?

The models are good enough

In the early days of building AI-powered applications, the models really had rough limits. Not that long ago we couldn't count on LLMs to reliably make tool calls.

There were no frameworks, patterns were not established, and prompting was a dark art.

We've made it through the bleeding edge and these models have gotten really smart and mostly reliable. The good news is you can account for most of the reliability issues easily with good code, the AI SDK, and good prompts.

This all has cut down on hallucinations dramatically all while LLMs have gotten much cheaper.

We're looking at a 100x price decrease between GPT-4.5 and GPT-5-nano (same 1400 ELO).

If it's fast, reliable, and affordable, there are more use cases you can build for. The limit is no longer the LLM; it is your ability to innovate.

New technology gives you an unfair edge

Sales is competitive by definition. Over time, all advantages are competed away. But right now, there's a gap. I've listened to some leaders at Fortune 500s talk about how their org is using AI. Cursor is amazing, ChatGPT is great — but if you are stopping there you are leaving a lot of money on the table.

We only have a small narrow window before everyone learns the trick.

Tim Ferriss, Gary Vaynerchuk, and Seth Godin all talk about the glory days of a Google ad being 5 cents a click. This is that moment but maybe bigger.

And it won't just help you get more customers; building better tools leads to better culture.

By giving employees leverage, leaning into the future, and removing grunt work, you're opening up time for your team to work on more difficult problems. Teams that regularly engage difficult problems well build better businesses.

Building internal tools has never been a better ROI

v0, Cursor, Claude Code, Codex: so many great tools that dramatically reduce the cost of writing and maintaining software. And the good news on internal tools, they're often greenfield projects where design doesn't matter. What a sweet spot for the current state of AI!

These LLMs have a lot of value in them. Buying off the shelf agents might make sense, but they're using the same models under the hood most likely. Fine tuning hasn't really been necessary to solve GTM use cases since the core LLMs keep improving so rapidly.

There are some niceties that buying solves but you'll still have to do some work to integrate it with your data. And even then, there's a good chance these tools do not connect to the data you need.

Here's a quick thought experiment: say you were going to buy a nice dress and you had a set budget. To stay within your budget, you considered a cheaper dress that somewhat fit.

Now imagine a magical tailor who discovered a 10x more efficient way to make bespoke clothing that fits perfectly. Because of the 10x improvement in efficiency, the tailor sets their price exactly equal to the off-the-shelf clothing.

Which dress are you picking?

Bespoke is always better, especially if the cost is equal.

How to get started

Starting a GTM Engineering team is simple but not easy. You need to find great engineers that are interested in the domain they're augmenting. You need to find the top 5%, the killers, and shadow them regularly. And then it's critical to prioritize ruthlessly.

Finding engineers

There are two types of folks that make great GTM Engineers:

1. Engineers that have seriously debated becoming a product manager
2. Sales/Solutions/Field engineers that like to build

It's critical you're able to get to the crux of "what would actually help us win this deal" or "how good is this email copy" or "how would I check if this was a healthy customer".

All of these questions are hard to answer if you don't have the PM itch or have already built up the sales and people skills necessary to answer these.

A mix of internal and external hires to build out your team is critical.

Shadow the killers

Your GTM Eng team will come up with some ideas, but the best ideas will come from watching your top performers. They're doing things no one else in the org is even thinking about and some of it is brutally time consuming.

Part of doing well in sales has always been tied to an ability to grind.

Find out what that differentiating task is & find a way to bring it to the rest of the team. You probably won't get them to a top 5% output, but you can get them to the 75th percentile and that's enough to change your entire business.

Problem selection is key

AI really is a generational technology. I am not counting on AGI when I say that (read: AGI: Some Assembly Required). Simply taking the existing technology and applying it to the entire economy is sufficient to keep our 2-4% GDP growth humming along for the next 50 years.

Since AI is so transformative, there are a lot of problems out there that could not be solved previously that are now solvable.

That does not mean you need to solve them.

Be picky. Your goal with GTM Eng is not to ship the entire kitchen sink, but to focus and find the key leverage points where the tech is currently good enough and the opportunity is large enough.

There are 10x returns to be had here, just don't get distracted by the 10-20% improvement ideas.

Get in touch

If any of this sounds interesting, there are two sets of folks I'd love to talk with:

1. Engineers that fall into the buckets I mentioned in the "finding engineers" section. I'm hiring! Apply and then email me to say hello.
2. Leaders looking to build GTM Eng teams. We're building in public because we think Vercel is the best place to power your internal agents and apps. Let's connect and get your GTM Eng effort off the ground in Q4 so you can hit the ground running in 2026.

Email address: drew at vercel dot com

If either of these are you or someone you know, please reach out!

Let's build the future of GTM together.