I have a confession. I hired a GTM engineer. Then I spent the next year building software so that most companies would never have to.
Warmly hired one in May 2025. We now sell Forward Deployed GTM Engineer services for $10-12K a year. And our core positioning? "You don't need to hire one."
That's not a contradiction. That's the actual state of GTM right now.
Last week, I was on an internal call where our own guy, Aleksandar, said it out loud: "What we're aiming for is to get the GTM engineer role out of the way and have sales directors and marketing directors use the technology as simple as possible. Instead of them having to build these clay tables and think about APIs and think about how do we send this and how do we pull this... it's a lot of work."
He's building the thing that replaces his own title. And I'm funding it.
The punchline? I'm also doing the job. I run product and marketing. I write the blog posts, build the landing pages, run the ads, manage the email sequences, and inform product decisions. The line between this role and marketing leader just disappeared.
The need is real. But who actually needs one, what they should be doing, and where the role is going? Almost everyone gets that wrong.
I occupy a weird position here. I'm an employer, a service provider, AND a builder of software that replaces the need. That paradox gives me a perspective nobody else has.
What a GTM Engineer Actually Does (Not What Clay Tells You)
The Old Definition: Clay List Builder
The market thinks the role = person who builds Clay tables, runs enrichment waterfalls, sends cold email. That's the 2024 definition.
Clay invented the category. They created the job title, built the community, ran a bootcamp, hosted a World Cup. Now every person with this title is a Clay user by default. And honestly, they built something powerful. When Brendan, evaluating tools for Datagrail, looked at the market: "Clay is 100% customizable... I need this level of customizability to do what I do."
He's right. For him.
But if the job is "manage Clay tables and send cold emails," you hired a tool operator.
The Real Definition: Full-Stack Marketing Infrastructure
Look, I know what people picture when they hear "GTM engineer." Someone hunched over Clay, dragging enrichment waterfalls, tweaking email sequences. That's maybe 20% of the actual job.
The real job is connecting everything. SEO, paid ads, email, LinkedIn, landing pages, content, retargeting, analytics, CRM, enrichment, attribution. All of it flowing into one system. All of it feeding back into itself.
The goal is to build the system that allows AI to see as much and do as much as possible, whether by itself or through people.
Build the infrastructure so well that it runs itself. That's the job.
This Is What My Week Actually Looks Like
Nobody writes this part. Every blog post about this role reads like a job description. "Manages data pipelines. Builds enrichment workflows. Coordinates cross-functional teams." Come on.
This is what I actually do. Every week. As one person running product AND marketing at a Series B company.
1. Find the gaps
Monday morning. I'm in Google Search Console looking at what keywords drive traffic. I see a competitor ranking for a term we should own. So I pull up SEMrush, cross-reference with Ahrefs, and prompt Claude Code to analyze the gap.
"GTM engineer" gets 1,900 searches a month. Clay owns it right now. This blog post is me taking it.
That's where the work starts. Not with a list of contacts. With a map of where the demand already exists.
2. Create the content
I write the blog post. I build the landing page. I record video content for social. I create playbooks from call transcripts.
I told my marketing team: "Copy the transcript, paste it into a new Claude Code session, and just say generate me a new playbook." Twenty minutes later, it's done. That's the speed we operate at.
3. Drive traffic
Google Ads pointing to the landing pages. LinkedIn ad audiences built from our TAM data. Meta ads. YouTube pre-roll. Retargeting across every channel where our buyers spend time.
From one system, I can target by persona and push to ads automatically.
4. Capture and identify
Warmly identifies which companies and contacts visit which pages. I can see their buyer journey. What content they consumed, how long they spent, what signals they're throwing off.
This is where most GTM stacks break. They can send. They can't see. We can do both.
5. Route and nurture
In-market accounts go to reps immediately. Not just "this company visited your site." Full context: what pages, how many people, what intent signals, what the buying committee looks like, what they should say in the first email.
Not-in-market accounts get automated sequences via Customer.io. Personalized, triggered by behavior. Not batch-and-blast.
6. Tune the machine
Track which content converts. Double down on winners. Kill losers. Shift budget to what's working.
I use LLM-as-a-judge on top of the full buyer journey to figure out attribution. I don't think anyone else does it this way. But it works.
7. Generate creative at scale
AI-generate ad creatives for LinkedIn, Meta, Instagram, YouTube, TikTok, X. Work with designers on refinement. Test variations. Kill underperformers fast.
8. Feed it all back
Every interaction, every outcome, every decision goes back into the context graph. The AI gets smarter. The next cycle is better than the last. It compounds.
I do all of this. I'm one person. That's the point.
Three months ago, our pipeline was $500K. Last month, $1.4 million. This month, we're on track to triple again. All demand gen. All driven by this infrastructure.
Shanzey on my marketing team said it well: "At my previous company, the marketing system involved so many people and so many systems and nothing was really automated. Over here, just two or three people are running the show."
The GTM Engineer and the Marketing Leader Are the Same Person
This is the thing I keep coming back to.
A year ago, to do what I do now, you'd need a content marketer, a demand gen manager, a paid media buyer, and a GTM engineer. Four headcount minimum. Maybe five.
I fired those job descriptions and hired AI. Not because the work is less complex. Because the execution is instant.
At a Series A through C company, these two roles are converging. The marketing team can just be one person. I do the writing, figure out the topics, prompt Claude Code, see content gaps, write the posts, make videos, create playbooks, run the ads, manage email sequences, and inform product decisions.
The role used to be its own function. Now execution is trivial. The hard part is making the right decisions.
Once you define your ICP and personas, the system should automatically push. Trigger-based outreach. Queued sequences. The human decides WHAT to do. The AI decides HOW and WHEN.
What still needs a human: brand taste. Design quality. In-person relationships. Strategic intuition that data can't show. The "should we go after this market" call.
But the wiring? The orchestration? The day-to-day execution? That's infrastructure now. Not headcount.
Max, our CEO, said it at all-hands: "Everyone's going to get more productive. I think we won't need to hire as many people as we grow and scale because all of us will be even more efficient with AI."
He's right. And the person in this role is the one who makes that possible for the whole revenue team.
Building the GTM Brain
What the Brain Actually Is
Think about what happens when a target account visits your pricing page.
A dumb system sends a templated email. "Hey, saw you visited our site!"
Our system does something different. It checks: who else from that company visited this week? What content did they see? Are they in an active deal? Did they talk to our chatbot? What industry are they in? What have similar companies needed? It crawls through all of that context, compresses it into a plan, and then acts.
That's the GTM brain. The central repository that both your reps and your AI query before making any decision.
Carina, our co-founder, defined it: "Our context graph is being able to pull any context about a company or a contact based on their activity on their website, including chat, where they dropped off, and then being able to generate a personalized email sequence."
Every decision gets logged with full context: what the system knew, what it considered, what it chose, and what happened. I call these decision traces. They're how you audit an AI system. And how it learns from its own history.
I told my team: "The go-to-market brain is stuff you can't see. It's all underneath. But that is actually how we are going to win as a product."
I wrote about this in detail in Building Agents for GTM.
How the AI Actually Thinks
When we talked to Vishnu at LangChain about their own GTM agent, he described the same pattern we use: "Any time a lead comes in, the agent kicks off, looks at the lead, sees if it's someone worth reaching out to, looks at past conversations with that person or customer, and routes the lead and a set of emails to the right person."
The agent doesn't see everything at once. It walks through the context layer by layer until it has enough to make a decision. Then it compacts what it learned, creates a plan, and executes. All by itself.
This is why the infrastructure matters more than the automation. The automation sends emails. The infrastructure gives the AI the ability to actually think about who should get what, and why.
The Memory Bank
I wrote about this in How I Run GTM With Agents and went deeper in Memory is the Moat: context compounds. Workflows can be copied. Memory compounds.
Any competitor can replicate "if persona = VP Sales, send template A." That's a workflow. It's just rules.
But building the infrastructure that captures every interaction, compresses it into understanding, and learns from outcomes? That compounds. And it can't be copied.
Surface-level stuff? Aleksandar built it in a day. Anyone else can too. The context graph underneath is what actually matters.
The person doing this work in 2027 builds:
-
A unified understanding of everything happening with every account.
-
Infrastructure that coordinates multiple agents without collision. I think the hottest thing right now is probably these agent harnesses.
-
The Market Is Still Split. Pick Your Side.
I'll give Clay honest credit. They built something powerful.
Brendan at Datagrail was right: power users love power tools. Custom enrichment waterfalls, bespoke scoring logic, 15 stitched data sources. They want to build.
David Chase, a CMO, chose Warmly over Clay for the opposite reason: he doesn't want to hire someone for this role. He doesn't want to manage and maintain many different tools. He wants the thing to work.
And honestly? That's most of the market. A full-time hire costs $80K-$150K+. A Clay agency runs $80K+ per year. And that's before the Clay subscription, the enrichment credits, and the engineering time to maintain everything.
When I mapped out the legacy GTM stack for a typical Series B company, the number was $920K. The Warmly bundle? $440K. Save roughly 50%.
If you need custom enrichment waterfalls and bespoke scoring logic, Warmly isn't for you. Not yet. We're not as customizable as Clay. Our enrichment waterfall is solid but it's still catching up on edge cases. We lose deals over this. I know because I read every churn note.
But one of the problems with Clay is the burden of choice. Because you can do so much, you end up not knowing what you're supposed to do.
Our bet is that most companies don't want that level of customization. They want it to work.
Clay Created the Category. AI Is Redefining It.
There's a piece from Burn It Down Marketing called "The Job That Doesn't Exist: Inside Clay's GTM Engineer Playbook." I shared it with my team and called it a cautionary tale.
The core argument: Clay realized their product was hard to use. Instead of making it easier, they created an entire job category around the complexity. You need a dedicated person to operate it. And the company making the product is also the one training them through bootcamps.
When the tool vendor is also the one defining who you should hire to use the tool, ask who that arrangement really serves.
My prediction, shared with my exec team in January: GTM agencies and teams are going to move from Clay to Claude Code. It's starting.
My CRO's reaction: "Oh wow... okay we may be on to something."
I said it on an internal call: "Clay is in trouble."
Workflows break when conditions change. Reasoning adapts. An LLM with the right context doesn't need a workflow. It needs a spec: "Who we're targeting, what we know about them, what has worked before. Figure out the best action."
I don't personally love workflows. I just want this thing to do the job.
The skills that make someone great at this work today (data thinking, system design, understanding buyer behavior) transfer perfectly to the new world. The specific tool doesn't.
Richard Sutton's Bitter Lesson: don't encode domain knowledge into systems. Build infrastructure that lets AI learn. Every hardcoded rule in your Clay table is domain knowledge that will be obsoleted when models get good enough to figure it out themselves.
We could try to win a horse race, or we could try to build a Ferrari. The Ferrari goes 2 miles per hour right now. But one day it'll go 300.
Do You Actually Need a GTM Engineer?
No. You need software that works out of the box. Signal-based platforms that identify in-market accounts, route them to reps, and handle nurture automatically. Don't hire a person to wire tools together.
No. You don't have enough complexity to justify the role. Use a signal-based outbound tool and focus your hiring on closers.
Maybe. But consider whether the answer is a person to duct-tape your tools or consolidating to fewer tools that work together. Omari at ProjectWorks was evaluating us to consolidate Clay, Apollo, HubSpot chat, Usergems, and Lemlist into one platform. Sometimes the answer isn't more wiring. It's fewer wires.
Probably yes. At this scale you need someone building infrastructure, not just running plays. This is where the role creates real value.
This is exactly why we built Forward Deployed GTM Engineer services. $10-12K a year vs $100K+ for a full-time hire. Pack Digital signed this in February. They get the expertise without the headcount.
A 60-Day Playbook That Actually Works
If you're stepping into this role (or you just hired someone), here's what the first 60 days should look like. This is what I run at Warmly.
Week 1-2: Build the Context Store
Before any agent can do useful work, it needs context. Not scattered across 12 SaaS tools. Queryable. Structured. Already saved.
Pull everything into one place: CRM data, intent signals, enrichment data, outreach history, ad impressions. Connect all channels. Google Search Console, SEMrush, Google Ads, LinkedIn Ads, Meta Ads, Customer.io, your CRM.
PostgreSQL with good indexing. No graph database required. It's a Postgres database with all your systems feeding into it.
Week 3-4: Design the Two Buckets and Connect Every Channel
In-market (route to reps) or not-in-market (nurture). That's the entire funnel. Build the logic that sorts your entire TAM into these buckets every morning. Automatically.
Then connect every channel: email, LinkedIn ads, Meta ads, Google Ads, YouTube pre-roll, TikTok, landing pages, SEO content, retargeting. Every channel feeds signals back into the brain. Every channel gets activated based on what the brain knows.
Month 2+: Let Agents Execute, Tune the Specs
Once the infrastructure exists, execution becomes an agent problem. I have 3-10 agents running in parallel right now. Building lead lists. Adding contacts to LinkedIn ad audiences. Writing content. Analyzing attribution.
The person in this seat designs what the agents do, monitors the output, and tunes the specs. My hire didn't become unnecessary. He became the person who designs what the agents do. That's more valuable, not less.
FAQ
What is a GTM engineer?
A technical role at the intersection of RevOps, sales ops, and engineering. They build the infrastructure that generates pipeline: data enrichment, lead scoring, outbound automation, paid ads orchestration, content distribution, and multi-channel coordination. The person who designs the machine, not the person who operates it.
How much does a GTM engineer cost?
Full-time: $80K-$150K+ depending on experience and market. Outsourced through Clay agencies: $80K+/year. Forward Deployed services (like what Warmly offers): $10-12K/year. Many companies find that modern signal-based platforms eliminate the need entirely.
What tools does a GTM engineer use?
The modern stack: Google Search Console, SEMrush, and Ahrefs for SEO. Google Ads, LinkedIn Ads, and Meta Ads for paid media. Customer.io or HubSpot for email sequences. Warmly for signal detection and visitor identification. Claude Code for content generation and analysis. A CRM (HubSpot, Salesforce) for pipeline management. And increasingly, custom agent harnesses that coordinate all of these autonomously.
What's the difference between a GTM engineer and RevOps?
RevOps focuses on process, reporting, and CRM management. This role builds automated pipeline systems. RevOps designs the dashboard. The engineer builds the machine that feeds it. More technical, more focused on building new systems than maintaining existing ones.
Can one person run GTM with AI?
Yes. I run product and marketing at Warmly solo. Blog posts, landing pages, paid ads across Google, LinkedIn, Meta, YouTube, and TikTok, email sequences through Customer.io, the entire demand gen engine. $500K to $1.4M in pipeline in a month. Build the right infrastructure and AI handles execution while you make decisions.
What is a GTM brain?
The central repository connecting all your channels, decisions, and outcomes into one queryable system. It stores context about every account, logs every decision the AI makes (decision traces), and learns from outcomes. The difference between sending cold emails and running a coordinated, multi-channel revenue engine.
Do I need a GTM engineer or a GTM platform?
If you need deep customization and can invest $80K+, hire for the role. If you want results without managing another person or complex tool, use a platform that handles signal detection, routing, and outreach out of the box. Many companies start with a platform and add headcount when their needs get complex enough.
Will AI replace GTM engineers?
It will redefine the role. Today's version wires tools together and manages workflows. Tomorrow's designs AI agent systems, builds memory infrastructure, and writes the specs that agents execute. Tool operators become system architects. More valuable, not less. But only for the people who evolve with it.
The GTM engineer role is real. It's just bigger than anyone thinks.
I hired one. I built services around one. And I do the job myself every day with AI.
That's not a contradiction. That IS the GTM market in 2026.
The role and the marketing leader just merged. One person with full context, AI infrastructure, and the taste to know where to point it.
Most companies don't need to hire for this. They need software that does the job. The companies that do need someone? They need the kind that builds full-stack marketing infrastructure, not the kind that manages Clay tables.
They'll build the memory. The memory will build the pipeline.
See how Warmly replaces the need for a GTM engineer →
Or get a Forward Deployed GTM Engineer if you want the best of both →
Last updated: March 2026