Today’s Martec newsletter was more than timely. It hit exactly the spots I’ve been thinking about lately.
First things first, if you have not signed up for Scott Brinker’s chiefmartec newsletter yet, do so today. Meanwhile for reference, here’s the LinkedIn post about today’s newsletter: https://lnkd.in/p/g8XmQYhe
Now on to the topic of discussion.
Claude Code/Codex/Cursor – the general class of AI coding assistants has changed the speed of building software. It has been proven a few times through a lot of amazing use cases. And if you do not believe them, ask your own product development teams. For the most part, AI has boosted productivity for both Product Managers and Engineers. So much so that shipping features is now faster than ever before.
It would be something worth celebrating, right? Except – the entire GTM flow needs a bit of fine tuning, and a bit of rejigging. GTM needs the new and killer features to be brought into the forefront of the customers the users
Theory of Constraints is a 1980’s principle and relevant every day.
Here’s the system as it actually exists. Engineering builds something. GTM has to understand it. Sales has to be trained on it. Customers have to absorb it. Prospects have to update their priors about what you even do. That whole chain is the system. We just turbocharged one stage of it. The building stage. The rest of the chain, the human part, runs at roughly the same speed it always has. Attention is slow. Trust is slow. Rewiring a mental model that someone has been relying on for eighteen months is very slow. So the constraint moved. It didn’t disappear. And the new constraint is human bandwidth, which is finite and contested and not something you can throw compute at. Every release is a withdrawal from an attention account. Customers. The sales team. Partners. Analysts who’ve been categorizing you a certain way since 2023. Each one has to stop, update their understanding, and change how they work. That account doesn’t refill at sprint speed. It refills at human speed, which is to say slowly, grudgingly, only when the value is obvious enough to justify the interruption.
So who actually owns the throughline from what gets built to what gets absorbed? That’s not a rhetorical question. In most companies right now, nobody does. Product decides what ships. Demand gen figures out how to promote it. Sales or Marketing gets a one-pager the week before launch. Customers find out via a changelog email they don’t read. That’s not a go-to-market motion. That’s a series of handoffs, each one losing fidelity, each one assuming the next team will somehow figure it out.
The teams that get this right won’t build less. They’ll be more ruthless about what deserves the cost of being absorbed. They’ll ask the constraint question first: is the bottleneck in engineering, or is it downstream in GTM, in the customer’s attention, in the sales team’s ability to explain the thing?
If the constraint is downstream, shipping faster just creates a longer queue at the bottleneck. And here’s where the product lifecycle comes in. If you can build a rough version of something in a week, you can also test whether it matters in a week. Prototype fast, put it in front of five customers, watch what they actually do, then kill it or double down before the full GTM machine even has to spin up. The goal is to move the constraint earlier, into discovery, where the cost of being wrong is low and the speed of AI gives you a genuine edge.
And bring Product Marketing and Sales sooner into the loop. Sooner as in, when you are thinking of the next killer feature or app. If you can build a rough version of something in a week, you can also test whether it matters in a week. Prototype fast, put it in front of five customers, watch what they actually do, then kill it or double down before the full GTM machine even has to spin up. Run it by a few people who are tuned to the customers’ needs. The goal is to move the constraint earlier, into discovery, where the cost of being wrong is low and the speed of AI gives you a genuine edge.
Eliyahu Goldratt’s insight was that the constraint is always somewhere. You find it, you fix it, then it moves. The work is identifying where it lives, not pretending it doesn’t exist. Right now, in software, it lives downstream of engineering. In the people who have to make sense of what you built.
Inspiration by Scott Brinker | LinkedIn and his chiefmartec newsletter. My opinions are my own and have no correlation to my employers – past or present. I use Huffl to structure my thoughts. Huffl is a multi-model, multi-modal canvas ideally suited for knowledge workers and students and works great for ADHD+ people like me.
Leave a Reply