Andrew McGuire
You walk into a Monday standup and ask what the AI stack cost last month. You get four answers. None from the same dashboard. Two from people who didn't know the third tool existed.
Three Claude Pro subscriptions on three different cards. A Cursor seat the dev team is paying for that the SDR team uses for prospect research. An n8n workflow that runs every night and nobody can show you the YAML. Seven AI tools the company pays for. Four more that people pay for personally and expense. None of it ladders to a number on the revenue line. None of it has an owner.
That is the GTM operational stack in May 2026. It is not a tooling problem. It is a role problem.
Two GTM newsletters tried to solve the role problem in the last two weeks.
Sophie Buonassisi at GTMnow named the Agent Operator on Friday. Define, deploy, evaluate, optimize. The four-stage loop. The role that watches the road on the agents and catches the drift before customers do.
Brendan Short at The Signal has been writing the GTM Engineer for the past year. Wire the systems. Build the Clay tables. Engineer the pipeline plumbing. Real craft, measurable impact.
Both pieces are right. Both pieces describe half the role.
While the newsletters were defining halves, xAI was hiring the whole. The job is public on Greenhouse right now. Title: Head of GTM, Systems & Agents.
The preferred-skills section is what nobody has said out loud yet. It says, verbatim: “Hybrid of architect and operator — owns both vision and execution.”
That sentence is the role. Architect of the systems. Operator of the agents. One person. xAI is paying $200K to $400K for them. They are not hiring two people.
GTM Engineer is the systems half. Agent Operator is the agents half. Same role. Two names. The market is calling it two things because the newsletters write to slightly different audiences and the role is moving faster than the language. xAI's title is the cleanest version. Systems and Agents. The “and” is the whole point.
The mistake most teams make is hiring half. The systems person who cannot write an eval. The AI person who cannot read a Salesforce schema. Both exist. Neither is sufficient. The role that fixes the operational mess has both legs.
The reason the role exists is that nobody is holding the systems and the agents in the same head. The shadow stack is now the size of the real stack. Costs are not tracked. Outputs are not owned. Three months in, half the agents have drifted. Six months in, the team is paying for tools nobody uses and nobody can find the eval that justified them.
You do not fix that with another tool. You do not fix it with a CRO who tries to govern the stack from above. You fix it with one person who can architect the systems and operate the agents. One person who decides what gets rebuilt, what gets deprecated, and what gets new guardrails. That is the role.
Every agent the operator runs needs an input. Most teams point their agents at commodity data and wonder why output is commodity. The same enrichment vendors. The same intent signals. The same prompts. The agents work. The output is forgettable.
The Tier 1 input that compounds is not purchased. It is the founder.
When a founder publishes consistently, content engagement becomes the input layer. Agents act on the signal. Outbound runs through the founder's real domain on a warm list. The trust is built before the first message lands. Take Austin Hughes at Unify. 571 posts. $139M annualized pipeline from two people running signal-based plays. The signals start in the founder's feed. The pipeline is downstream of the founder showing up consistently.
Tier 2 and Tier 3 are different work. Volume is the gate. The right architecture there is signal-based outbound on a warmed sending stack, with hiring data, funding announcements, and stack changes as the input. Burner domains are not broken. They are the right tool for that tier. Same operator. Two loops. Right input on the right tier. The role spans both.
I have been running the integrated role for a year, before either newsletter named a piece of it.
Twelve months building one founder's LinkedIn by hand. Writing posts, editing for voice, tracking what landed, adjusting. One founder, 4M+ impressions in twelve months. Pipeline followed. Hires came through. Inbound deals showed up. Podcast invitations arrived that nobody applied for.
Then I built the agents to run the production. Five of them. Each one has a written task definition, an eval framework, and a weekly optimization cycle. Sophie's loop, applied to the founder as input. Brendan's systems thinking, applied to the stack underneath. Same person doing both because in practice the work does not separate cleanly.
The Tier 2/3 stack runs on separate sending infrastructure with its own architecture. Different inputs, different guardrails, different cost model. One operator deciding what runs where.
I make the judgment calls the agents cannot. What to publish. Who to contact. What is good enough and what gets rewritten. Nothing ships without a human deciding it should.
Sophie called the trajectory. The role is on the same curve RevOps was on in 2018. Niche title in 2017. Board-level conversation in 2019. Every Series B company by 2021. This role is moving faster.
The companies hiring for it right now are calling it different things. Head of GTM Systems and Agents at xAI. GTM AI and Innovation Manager at Notion. Director of GTM Innovation at Zapier. The titles vary. The work is the same. xAI's title is the most honest version of it.
Whatever your org calls it (GTM Engineer, Agent Operator, Head of Systems and Agents, RevOps lead with an AI mandate), it is one role. The companies that hire one person who can do both halves will compound. The companies that hire two will spend the year arguing about whose stack it is.
Both terms describe the same emerging role from different angles. GTM Engineer (named by Brendan Short at The Signal) emphasizes the systems half: wiring CRMs, automating workflows, building the pipeline infrastructure. Agent Operator (named by Sophie Buonassisi at GTMnow) emphasizes the agents half: defining, deploying, evaluating, and optimizing AI agents in production. xAI's job posting “Head of GTM, Systems and Agents” is the cleanest version of the title because it names both halves. The role is one job, not two.
xAI's Head of GTM, Systems & Agents owns the GTM tech stack (HubSpot, Clay, Gong, internal tools), architects and automates workflows, deploys AI agents that generate revenue, builds consumption-based forecasting, and develops dashboards for leadership visibility. The JD describes the role as a “hybrid of architect and operator” who owns both vision and execution. Salary band: $200,000 to $400,000.
Most GTM teams have a shadow AI stack the same size as their real one. Tools paid for on personal cards. Workflows nobody else can read. Agents running in production with no owner watching the output. Costs are not tracked. Drift is not caught. Six months in, the team is paying for tools nobody uses. Fragmentation is the operational pain that makes the integrated GTM Engineer plus Agent Operator role necessary.
The founder is the proprietary input that feeds the operator's loop on Tier 1 accounts. The founder publishes consistently. Engagement creates signal. Agents act on the signal. Outbound runs through the founder's real domain on a warm list. Trust is built before the first outbound message lands. Tier 2 and Tier 3 use different inputs (hiring data, funding signals, stack changes) on warmed sending infrastructure separate from the founder's primary domain.
A weekly letter. Free. Never spam.