AI in GTM
What AI actually changes in demand gen ops
AI doesn't build relationships or make the judgment calls. It removes the busywork between them, so the team can show up for more people at once.
What most teams get wrong
Most teams buy AI tools before they decide what work to hand over. They end up with a stack of pilots and no change in how pipeline gets built.
The better question is simpler: what work does the team repeat every week that a machine could do as well?
Where I learned it
In 2025, agentic AI was changing how marketing work gets done, and I wanted to learn it hands-on, not from a slide. So I built a consulting practice as a proof of concept and ran its go-to-market on AI.
That was version one. Twelve months later, I rebuilt it on the latest agents, and it's running my job search right now.
Coupa Cafe in Palo Alto, right before closing. My first client playbook was due in the morning, and the GPT started to hallucinate. The playbooks it wrote didn't map to my experience. So I started over and reworked it. Long night, Red Bull and pizza. But we made it.
AI writes fast. It doesn't know what you know.
Version one: the consulting engine
- ICP and list. AI helped me define the ICP for my services, then build and score a list of 100 CMOs I knew directly or through mutual connections.
- Research. AI researched each CMO before I wrote a word.
- Personalization. Every touch was one-to-one, across all 100 relationships.
- Hand-offs. Data moved between ChatGPT, Clay, HubSpot and Outlook without copy and paste. I checked each profile in LinkedIn Sales Navigator myself.
The result: 10 meetings and 2 signed clients from 100 CMOs.
To be fair, it was a warm list. AI didn't build those relationships. It let me show up for all 100 of them at once.
For two clients, I went past the coaching I was hired for. I built AI agents loaded with my expertise in partner marketing and campaigns, and turned their output into a how-to playbook for each new hire. One hire loaded it into their own GPT, named it after me and uses it as a virtual coach. The other still uses theirs 12 months later.
Version two: this job search
Hiring a marketing leader is a lot like a B2B buying decision. So I'm running my search the way I'd run demand gen, with an AI team built like a B2B marketing org: market intelligence, ABM, product marketing, SDR, content, enablement, marketing ops and RevOps.
A few rules keep it honest:
- Agents draft. I send. Nothing goes out, posts or publishes without my approval.
- One source of proof. Every result comes from one sheet. No agent can invent or round up a number.
- I own the relationships. Agents find warm paths. I decide who to call and what to say.
The funnel on my homepage is live and updated weekly.
Where a human still matters
- Knowing the people on the list, and which ones to call first.
- Judgment on positioning, pricing and what to say no to.
- Every conversation that follows.
I log every step that breaks or repeats by hand. After six weeks, the stable, high-volume steps move from agents to code.
What I'd do now
- List the repeatable work. Research, scoring, personalization and moving data between tools are the first candidates.
- Hand those to agents, with rules. Agents draft. Humans approve anything a customer sees.
- Keep people on the relationships. The time AI saves goes back to customers.
- Log the friction, then build. Move a step to code only when it's scheduled, high volume and stable.
Which part of your demand gen ops would you hand to an agent first?