How a $21M B2B SaaS CTO went from carrying AI pressure alone to presenting a board-approved plan his whole leadership team could execute.
This is what AI Catalyst looks like when the pressure is real.
The CTO had been reading about AI for months. Experimenting on weekends. Thinking about where it could help his product org. None of that was the problem.
The problem was that his CEO had already promised an AI strategy to the board. Three of his top 20 accounts were evaluating competitors who were marketing AI features aggressively. And his engineering team was at capacity, managing a 9-month product backlog with 30% frontend test coverage on the mobile platform that was supposed to be their competitive edge.
He couldn't translate any of what he knew about AI into a plan his VP of Engineering could execute without burning out the team, his VP of Product could scope, or his CEO could take to the board.
"All of my previous decisions and pitches have been from a place of certainty. I knew the technical approach, I could estimate the effort, I could predict the outcome. With AI, that just isn't an option. The landscape changes monthly. I kept waiting for things to settle enough to make a confident bet. They're not going to settle."
The company sells project management software to mid-size firms in a specialized vertical. They'd grown from zero to $21M ARR in four years by moving fast and capturing market share from legacy competitors. The growth strategy worked. But the cracks were showing.
Net revenue retention had slipped from 115% to 108%. Their best customers were hesitating to expand because they were waiting for AI-powered document intelligence the company hadn't built yet. Meanwhile, customer-impacting quality issues were hitting roughly every other release cycle. And the CTO was carrying all of this alone.
He didn't need an AI tutorial. He needed to make a defensible decision under uncertainty, get his leadership team behind it, and present something to the board that wasn't a slide deck full of buzzwords.
We ran AI Catalyst: roughly 30 days to co-create the plan, then 30 days of weekly coaching to pressure-test it in the real world. In this case, the plan came together in three weeks across four working sessions. The goal was not to hand him a plan. It was to help him build one he could defend.
We started with the business pressures the C-suite was actively talking about. It was immediately clear that the AI challenge was inseparable from broader problems. The retention gap wasn't just a product issue. Customers wanted capabilities the company couldn't build fast enough, and the bottleneck wasn't headcount. It was how they worked.
We mapped the political landscape: a CEO who'd promised AI to the board without consulting the CTO. A VP of Engineering pushing back on capacity. A VP of Sales forwarding competitive intel weekly, creating timeline pressure. The CTO was carrying all of this and hadn't articulated the connections between the pressures.
The CTO came back with hypotheses about where AI could create value. One was strong. Three were weak. That's normal.
Through collaborative pressure-testing, the hypotheses sharpened into two parallel workstreams: stabilizing the mobile platform using AI-assisted engineering tools (protecting the competitive advantage), and running prototype-driven customer discovery on AI-powered document extraction in the field.
The CTO had been thinking about these as separate initiatives. They were actually two chapters of one story, converging into the company's biggest market opportunity. Protecting the core while validating the next bet.
"You didn't give me the answers. You kept asking me questions until I found them myself. Which is annoying and effective."
The CTO pressure-tested the plan with his VP of Engineering, VP of Product, and head of Customer Success. The VP of Engineering was more excited than expected. The CS team provided a concrete baseline of customer complaints to measure against. The VP of Product saw how prototype discovery could transform her team from backlog managers into leading the effort to create better products.
We built the playbook, rehearsed the board presentation through role-play, and prepared messaging for three audiences: the CEO and board, peer leaders, and the product and engineering team.
The CTO was leading with the feature instead of the strategy. The board needed to hear "we have a repeatable system for AI investment" rather than "we're building document extraction." Once he flipped that framing, everything clicked.
"Setting the target around reducing customer impact rather than just hitting a coverage number helped the team prioritize the tricky areas of the application. The parts that actually caused problems in production. That distinction made the whole initiative feel more purposeful."
8 customer discovery sessions completed with accounts selected alongside CS. Key finding: submittals were significantly more important to customers than RFIs for extraction, reshaping the build priority entirely.
Accuracy threshold clarified: ~90% extraction accuracy was sufficient (not the >95% the team assumed) as long as corrections were easy to make inside the application without switching tools. One test subject suggested the tool highlight fields it wasn't confident about. The next subject was asked about this and responded with visible enthusiasm. This became a core design requirement that would not have emerged from internal planning.
"We went from guessing what customers needed to watching them use the prototype and learning things we never would have figured out on our own. The team is talking about being more customer-focused and less focused on backlog management. That's a culture shift, not just a process change."
First informal 90-day planning cycle completed with a green/yellow/red capacity framework. AI Slack channel hit 47 posts in 90 days from 18 different team members. Two new AI experiments surfaced as candidates for the next cycle. The CTO presented the AI strategy to the board at week 6. Board approved continued investment.
"They helped me reframe struggles I'd been having in a way that let me be bolder and take business risks without making them personal risks. Having a structure for that changed everything."
The CEO told the CTO afterward that it was the strongest technical presentation in recent memory.
Rather than a comprehensive AI transformation plan, the team adopted a 90-day cycle for evaluating and executing AI investments. Each cycle produces focused bets with 30-day signals and 90-day decision points. The board doesn't face make-or-break decisions. The investment stays small enough to generate real ROI quickly.
The CTO built the plan. We provided the structure, pressure-testing, and suggested framing. Every hypothesis, every bet, every narrative choice was his. That's why he could defend it to a skeptical board without slides and adjust in real time when challenged.
The technical plan was necessary but not sufficient. The CTO also needed to pre-wire his CEO, activate his VP of Sales as a competitive intelligence asset rather than a source of timeline pressure, and sequence communication so nobody was surprised.
The team co-created AI adoption principles that democratized decision-making. Instead of every AI question flowing to the CTO, the principles gave people a framework for making good decisions independently.
"The principles aren't rules I'm imposing. They're the operating system we're agreeing to run on."
The company is entering its second 90-day cycle. The document extraction build is scoped based on validated customer requirements. The 90-day planning cadence is being formalized. Collaboration around leveraging AI is now done in the open. Practices are shared, as are conversations about risk.
The CTO now has a system, not just a plan. The plan will change. The system won't.
The point wasn't that everything got easier. It was that the team finally had a way to make hard calls without pretending they had certainty.
30 minutes. No pitch. We'll tell you if Catalyst is the wrong fit.
Request a Fit Call →* The account renewal is correlation, not causation. However, the mobile stabilization hypothesis was the CTO's chosen focus area, and other anecdotes suggest the work has been noticed and appreciated by clients. Company name anonymized.