AI Did Not Change GTM. It Exposed What Was Already Broken.
Fix your broken GTM architecture and build compounding growth loops with Jacco van der Kooij.
This newsletter is kindly supported by Attio - the AI-native CRM built for modern GTM teams. If you’ve ever battled a rigid legacy CRM, Attio fixes that with clean data, real-time visibility, and workflows that actually speed you up.
Today, 5,000+ fast-moving companies rely on Attio’s flexible data model, modern design, and builder-friendly setup to shape their CRM around how they work - not the other way around. It’s built for founders, revenue ops, and GTM teams who need a system that adapts as fast as we do.
So if 2026 is the year you finally break up with your legacy CRM or start fresh on a new project, this is the perfect moment to take Attio for a spin.
Dear GTM Strategist!
Before we jump into another exciting round of AI playbooks, we need to talk about the infrastructure we are operating in. Most conversations about AI in go-to-market focus on tools and tactics. How to automate outreach. How to build agents. How to scale workflows.
These conversations are exciting, but they often miss the bigger issue.
Artificial intelligence did not fundamentally change GTM.
It exposed how many of our systems were already broken.
For years we built revenue engines that relied heavily on human effort. People writing cold emails. People updating CRM records. People manually interpreting customer feedback. It worked when capital was cheap and headcount was easy to justify.
Those systems are now under immense pressure.
Jacco van der Kooij, the brilliant GTM strategist leading the Winning by Design movement, frames the shift in a very intriguing way.
“For years, we talked about keeping the human in the loop. Today the real challenge is to close the loop.”
Closing the loop means building systems where learning happens continuously. Insights move quickly through the organization. Decisions are based on structured data rather than fragmented opinions.
That requires something many GTM teams would rather avoid. It requires looking at the infrastructure behind how revenue actually works.
After reading Jacco’s chapter in Momentum’s Ignite Your GTM with AI, I spent months trying to get him to talk about this topic with us. If AI is going to work inside GTM systems, we first need to understand the architecture those systems run on.
And that architecture is often messy.
Teams are building AI workflows on top of fragmented data, disconnected tools, and processes that were never designed to scale. Instead of creating leverage, this often creates more complexity.
This conversation is about fixing that foundation.
It is not a light read. But if you are responsible for leading GTM teams in the AI era, it will change how you think about your systems.
In this edition, we’ll cover:
Why three “black swan” events colliding at once created the perfect storm for GTM
How AI-native companies build compounding revenue machines instead of linear sales motions
The four data crises you must fix before AI can actually help you
Why the excavator analogy should change how you think about your team
How to optimize for buying instead of selling, and why that distinction matters more than ever
Let’s get into it.
The Macro Shift and the AI Revenue Machine
Artificial intelligence did not change go-to-market overnight. It simply exposed what was already broken. And this massive shift did not happen in a vacuum. Jacco points out that it is the direct result of what he calls the Three Black Swan Convergence - three compounding macro events colliding at the exact same time.
COVID-19 fundamentally changed how we work and interact. The sudden end of zero interest rates completely changed how we fund businesses. And artificial intelligence changed how we build and scale products. Each of these on its own would be disruptive. Together, they exposed a massive systemic flaw in how software companies operate: for the last decade, we over-indexed on cheap human labor instead of building truly intelligent systems.
When money was practically free, founders could raise millions on a mediocre idea and hire armies of salespeople to brute-force their way to revenue. We bloated our recurring revenue models with manual, unstructured work. Humans writing cold emails. Humans manually updating CRM records and parsing through customer feedback. Every single time you insert a human into a repetitive data process, you introduce delay, bias, and error. For years we compensated for these slow feedback loops with more effort and more headcount. The system was highly inefficient, but it was still moving. Now that inefficiency is completely visible.
Jacco explained this perfectly:
“The main change that we see over the last two years is that AI has revealed how recurring revenue is supposed to work. And it did that by removing the influence of people in an unstructured way, that it actually slows the system down.”
AI-native companies operate entirely differently. They strip away unstructured human influence and build closed-loop systems. If a new user segment starts converting faster today, the system detects it immediately and automatically routes more resources to that specific segment. Learning takes minutes or hours, not months.
This speed is driven by what I’d call User-Decider Convergence. The old software model relied on top-down procurement - you sold to the executive, and they forced their team to use the tool over a nine-month sales cycle. AI natives flip this entirely. The person using the product is also the champion and the decider. They experience the pain, find the solution, and feel the impact instantly. They swipe their own credit card. Adoption spreads organically from the bottom up, and then they upgrade their own account and invite their entire team to join.
This is how you build a compounding revenue machine. This is how a company scales from $1M to $100M without hiring a thousand salespeople. You don’t just buy a new AI tool - you change your entire operational architecture.
The Four Data Crises
Before AI can act as your growth engine, you must fix your data infrastructure. You cannot scale a broken system, and if your underlying data is a mess, adding artificial intelligence will only help you make expensive mistakes much faster. Most traditional software companies are currently paralyzed by their own tech stacks, trying to deploy advanced AI models on top of a broken foundation.
Here are the four critical data crises you have to solve first:
Unstructured Data. Ninety-five percent of your customer conversations remain unformatted and completely unused. Your sales team has hundreds of calls every single week, and they take highly subjective notes based on gut feelings. This unstructured data is completely useless to a machine - AI cannot parse a vibe. You must use a strict methodology like the SPICED framework to turn human conversations into structured, machine-readable data.
Incompatible Data. Your tools do not speak the same language. Sales call recordings live in one platform. Pipeline revenue lives in your CRM. Customer support tickets are stuck in a completely different system. They exist in isolated silos, and you cannot build a unified growth engine if your data is fragmented across six different dashboards that refuse to integrate.
Disconnected Data. This is exactly what breaks the compounding growth loop. A customer success manager might learn exactly why a major client churned on a Tuesday, but that critical insight never flows back to the marketing team running the acquisition campaigns. The system is entirely disconnected - the right hand has no idea what the left hand is learning.
Incomplete Data. Gaps in your data capture are incredibly dangerous in the AI era. If you feed an AI model incomplete data, it will not just misfire - it will actively mislead you. It will confidently tell you to target the exact wrong segment and burn your entire marketing budget in the process.
To fix these four crises, you need to build a real-time data infrastructure. This is the core difference between the old way of operating and the new AI-native approach: you need a closed-loop system where data flows continuously. Learning from a customer cannot take an entire quarter. It must take hours.
Here is how Jacco van der Kooij explained this urgency:
“What do I mean with real time? Well, look, if you learn something happens in a region or if you learn a particular conversion rate is in a particular state and it moves down. It cannot take months before that propagates and leads to action. It is got to take minutes, hours.”
If a conversion rate drops abruptly, your system needs to adjust that same afternoon. If a new and highly profitable buyer persona suddenly appears in your funnel, you need to throw more resources at that fire immediately. You can no longer afford to wait for a monthly reporting meeting to figure out what the market wants.
The Excavator Analogy and the 30/70 Talent Split
We are currently forcing our most expensive talent to do manual labor. We pay human beings to write generic outreach messages, scrape lead lists, and update pipeline stages. This makes absolutely no sense.
Jacco uses a brilliant construction analogy to explain this shift. If you need to dig a massive hole, you can hire four people with shovels - they will eventually get the job done. Or you can bring in an excavator that digs faster, deeper, and with far more accuracy. AI is the excavator. But here’s the important nuance: when you bring the excavator onto the site, the humans don’t just disappear. Their roles completely change. You suddenly need an inspector to make sure the machine doesn’t hit a gas line. You need an architect to verify the foundation is square.
This is how Jacco explains it:
“The human beings are taking away the role of physical labor, the shovel, and are now becoming humans that oversee if the work is done properly.”
Your team should not be writing the spam. They should be managing the AI agent that generates the campaign - reviewing the target list, tweaking the context of the messaging, approving the final output. Once they perfect that process, they build an AI agent to check the first AI agent. This is a fundamentally different way of working, and it requires fundamentally different skills.
This shift brings us to a harsh reality about talent. AI will not fix your mediocre performers. Most leaders assume that AI will magically elevate the bottom 70% of their team - that it will turn struggling reps into top performers. It will not. If a rep doesn’t understand the buyer’s pain points, giving them an AI tool just helps them send irrelevant messages much faster.
The real magic happens at the top. AI gives your top 30% superpowers. These are your relationship builders, the complex problem solvers who actually understand the market. When you remove their manual data entry and administrative tasks, they operate at unprecedented levels. So stop trying to use AI to save the bottom 70%. Instead, architect systems that eliminate the need for routine execution work entirely.
Avoiding the Sales Tech Sidestep
Sales technology took a massive wrong turn over the last decade. The earliest tools were built to help us use data to work smarter - they were optimized for intelligence. Then we took a detour. Later tools completely abandoned intelligence and optimized for pure volume instead. We built massive machines to automate cold outreach and scale email spam. AI is now forcing us to return to that original principle: using data to reach the right people at the right time with genuinely valuable information.
This is where early-stage founders have an unfair advantage. If your company is under $5M in revenue, you are in the best possible position. You don’t have to unlearn bad habits. You don’t have a massive legacy tech stack to dismantle. You can build a clean AI-native system from scratch.
Jacco shared the exact tactical workflow with me, and it is incredibly simple:
Use LinkedIn Sales Navigator. Set strict filters to pull a highly targeted list of 200 prospects.
Export that list and feed the profiles into an AI model like Claude. Ask the AI to analyze those 200 people and find the exact common problem they all share right now.
Use a tool like HeyReach to build a highly contextual campaign based on that specific research.
A few years ago, this level of research and execution would take a team of human SDRs several days to complete. Today, a single founder can do the entire thing in three hours. You don’t need a massive team to compete anymore - you just need a smarter architecture.
Optimize for Buying
We need to talk about buyer psychology, because this is where everything connects. People absolutely hate being sold to - but they love to buy. If you stop someone on the street to aggressively sell them a timeshare, they will run away. If you put that same person on Amazon, they will happily spend hours filling their cart. The motivation is completely different when the buyer is in control.
“Most people hate being sold but they love to go buy. And that means that the sales process should not be based around selling and optimizing selling; it should be based on buying and optimizing buying.”
The old software model completely misunderstood this dynamic. We built massive, complex processes optimized entirely for the seller - mandatory discovery calls, artificial demo gates, aggressive outbound follow-ups. We tried to control the timeline. The new AI-native model uses technology to optimize the buying process instead. It removes friction. It uses AI to figure out exactly what specific research or proof the customer needs to make a purchasing decision today. It doesn’t blast them with templates - it finds the patterns in their problems and offers the exact solution they are actively looking for.
This is where true scale happens. When you optimize for buying, you activate compound growth loops. You achieve exponential growth when your current user essentially sells your product to the next prospect. They experience a frictionless buying process, get immediate value, share the tool with their internal team, and invite external collaborators. The adoption loop feeds itself, and AI simply accelerates how fast that loop spins.
Value Versus Impact
The final piece of the AI revenue machine is how you communicate, and most B2B companies get this completely wrong. They all sound exactly the same - saving time, reducing costs, boosting productivity. They focus entirely on the value of their product, but value is simply what your product can do. It’s a feature. If you sell a car, the value is that it drives fast or uses electric energy. The impact is entirely different: the impact is that the car allows you to visit your mother who lives a hundred miles away.
Jacco summarized this dynamic:
“The impact is an outcome, the value is a promise. We often say value is the promise of impact. Impact is the realization of that promise.”
When you sell value, you’re just making a promise. Buyers don’t care about your promise - they care about their own reality. This means you have to completely shift your messaging. Look at what your early customers actually achieve with your product. Listen to the exact words they use to describe their experience.
And here’s where most marketing teams ruin everything: they take raw, powerful customer feedback and try to make it sound professional. They sanitize it. They polish it until it sounds “clean and sexy.” When you do that, you lose the touch - you lose the exact verbiage that actually resonates with your buyer. Don’t polish your copy until it sounds like every other software company. Use the exact words your early customers use in your marketing campaigns. Keep it in their natural language.
When you are trying to differentiate your brand in a noisy market, you will have to make hard choices about which pain point to attack and which messaging to scale. This is where the human element becomes your ultimate advantage in an AI world. You can’t just wait for a dashboard to tell you what to do - you have to rely on your intuition to find the starting point.
Jacco van der Kooij left me with one final piece of advice that I think captures the whole conversation beautifully:
“Your gut tells you where to look for the data, then the data will narrow it down and make sure that you get it right.”
Trust your gut to find the initial direction. Then use your data architecture to prove it’s actually right. This is how you win the next era of go-to-market.
Live next week: Clay for Growth & Marketing
Next week, I’m hosting a live stream you won’t want to miss.
On March 19th, I’m joined by Davide Grieco, Head of Growth at Clay, to walk through how growth and marketing teams are using Clay to move faster, act on signals in real time, and finally get credit for the ROI they generate.
We’ll demo real playbooks - from auto-refreshing ad audiences and signal-based campaigns to personalized landing pages at scale - the exact workflows teams at OpenAI, Vanta, and Verkada are running today.
If you’re tired of slow campaigns and budget justification loops, this one’s for you.
Launching a product is hard.
You do not have to guess your next steps.
I built the ultimate 100-Step Go-To-Market Checklist to help you launch with total confidence and zero chaos. It includes proven website templates, sales decks, landing pages, outbound sequences, LinkedIn post frameworks, email sequences, and 20+ workshops you can immediately run with your team.
It covers every single phase of your go-to-market motion.
→ Customer research and positioning
→ Pricing and packaging strategies
→ Distribution and growth loops
This is the exact step-by-step framework I use to help my clients scale.
Stop guessing and start executing.