What a month of selling taught me about AI GTM
I've been out there selling in earnest for a month. A lot of calls with prospective buyers, and some with advisers on how you take this kind of work to market.
We chose to work with large B2B companies. This month we could have chosen to work with Series A and Series B companies instead. I dug my heels in. I think the thesis is still right. We're not giving up on our segment. We have to make our GTM good enough to meet it.
The buyers are being asked what their AI strategy is and how to resource it. Everything is chaos right now. Do I buy this product or that product or do I get everyone Claude licenses?
My read is that most companies run AI GTM as a tool rollout. Nobody owns it as a product, so there's no loop. Individuals get faster and the number doesn't move.
Customers evaluate us based on how they buy from us. I think AI will take out many of the handoffs SaaS built. The buyer should get a useful answer at the first touch, instead of waiting for the next person.
You can start with one bounded team. Give it an accountable owner, a business goal and a definition of success. Put a builder on it who has lived the work, and run the whole loop there.
At the company level, that takes product leadership beside the CMO, with authority across marketing and sales. Next year's revenue plan should set its goals.
Builders change the work. The owner keeps checking the result against the goal. What happened changes what gets built next. When an improvement works, it reaches the whole team.
Buying more tools leaves those decisions unmade.
There's all these people talking about context. Shared context matters, but it sits underneath this work. It can't choose the goal or decide whether the result was good enough.
What I care about is the improvement that survives the person who made it.
The value comes from compounding improvements and treating this thing as something that we love that gets better over time.
We call it GTM product thinking. The talk walks through the shared foundation and the work it makes possible.
Part Run AI GTM as a product loop
What a product loop runs on
The talk starts with the shared harness, every place AI touches your team's GTM work. Shared context sits under the harness, and the harness sits under the product loop.
One meeting-prep skill for every seller
On my own, I can fix a skill for my next meeting. A team needs more than that.
I'd have a product thinker gather what sellers need from meeting prep and curate one shared skill. Sellers use it and report what helped or what was missing. The owner checks whether the prep meets the agreed standard and uses that result to choose the next change. The improved skill goes back to everybody.
Sellers can still build. The curation has to happen somewhere.
The talk illustrates the sharing with a sales-deck skill. One instruction changes so every deck starts with the buyer's problem, and the whole team gets that version when they next update their skills.
Read more about shared improvements in the Field Guide.
The buyer should get an answer at the first touch
Most GTM teams follow SaaS GTM design. Roles are set by what one person can do. Process follows the specialization, and software is sold by the seat.
Every role exists because the buyer needed something. The buyer climbs over every handoff to get it. Why should a buyer wait through qualification and another handoff before reaching someone who can show a useful demo?
I think AI can compress that sequence.
Amanda Kahlow describes the destination:
…the buyer gets expert depth at the first touch instead of four handoffs later.
Mark Roberge spells out the cost of specialization:
With specialization, you can hire the perfect person. But now you have these handoffs that are very inefficient for the customer… AI can empower a single individual to handle all those roles.
He adds: “It doesn't mean you're getting rid of those people.”
The point is fewer handoffs for the buyer, not fewer job titles.
Who decides which skills your sellers should rely on?
In the talk I ask Claude for a skill that logs my Sculpt meetings to HubSpot. I run it on one meeting and check the record in HubSpot.
GTM needs a mix of build fast and build right.
Build fast gets the first skill running. Build right includes deciding whether the team should rely on it.
That decision is still open after the demo works. The Field Guide entry on GTM product thinking covers who makes it.
The revenue plan sets the loop's goals
The goals come first. A product leader beside the CMO reads next year's revenue plan and identifies the bottlenecks that put it at risk. They choose which one to address and what result would count as success.
I use this map to connect improvements to today's tasks with changes to the process itself.
The picture shows the mechanics, with the harness and shared context underneath. The goals and the judgment still need an owner.
Then the map shows who does the work. The builders, your GTM engineers and RevOps, sit between the team and the systems.
The team tells them what it needs. The builders return improvements as shared skills and tools. Mostly, the team asks for today's job with better tools, and that's worth having.
Build right asks whether the process itself should change. The map includes work that can run without a person at each step. Builders still need a leader who owns that process to choose the change with them.
Two builders have described the same limit to me. They could make a bad process less painful with a skill. They needed someone to rethink how ideas got onto the web. We wrote the builders' side of this down in the Indie GTM Engineer Manifesto.
How are you going to compete next year if your web launches take six months?
Every box is a place to ask: what's the bottleneck, and who fixes it?
Three workflows the whole team can use
I run three workflows from our own shared setup.
Why do BDRs alone carry event invites, when everyone on the team has a first-degree network? I want that capability available to the people who can help the buyer.
- Event invitesI invite people to our Sculpt meetings with the tracking a BDR program used to hold.
- Studio and BrandI pick an image from our Studio gallery and save it into the page I'm editing.
- ABM campaignsClaude swaps an existing figure into a live ABM page, then I check the page after the rebuild.
These demos show capability the team shares. They don't show it working at your size.
Check how we run ours
We say we run our own GTM this way. The breakdown is where you check: the log of this campaign, what we set out to do, the calls I made, what the agents built, what broke, and what we changed.
How we built this campaign, and what broke
To argue with it in person, book time with me.