AI won’t save your strategy. Imagination will

“Sean,” the CEO of a health technology startup, invited Garry to join what he called a strategy offsite. By mid-morning, it had become something else. Six people, a whiteboard, a Series A closing in ninety days. The question on the table: where do we invest next? More engineers? A bigger sales team? Another content hire?

Garry set down his coffee and asked a different question. “What if we stopped asking what we can afford to build, and started asking what we can now imagine building—that we genuinely couldn’t have a year ago?”

The room shifted. So did the conversation. And that reframe—from resource allocation to possibility expansion—is the strategic inflection point that most leadership teams have not yet reached, and it is the one that will separate the companies that use AI to compete from the ones that use it to lead.

The research reinforces this point. A recent report from Deloitte finds only 34% of organizations are using AI to deeply transform their work. Another report from McKinsey confirms that companies using AI to pursue growth are 3.5x more likely to achieve enterprise-level impact than those focused on cost reduction.

Drawing on our work with senior leaders—Jenny as an executive coach and leadership development expert, and Garry as a physician-entrepreneur who mentors and advises early-stage startups—we have identified three strategies for leading innovation in the age of AI.

1. Start with the Friction, Not the Platform

When Garry first met with the founding team of “Axiom Health,” a startup focused on next-generation cardiometabolic biomarker analytics for early disease detection, their conversations were almost entirely about technology. The platform was front and center; the patient was not.

He pushed them to slow down. Before talking about what to build, he asked them to map the real tension their future patients and physicians were living with: Where were clinicians hitting walls? What workarounds revealed unmet needs? And what would need to be true for a better solution to be trusted at scale?

This is the discipline Marilee Adams calls “question thinking” in her classic business book Change Your Questions, Change Your Life: the shift from judger questions to learner questions. In an era where AI accelerates execution, the quality of the question matters more, not less.

Ask yourself:

• What are the top ten friction points our customers experience—including the ones they’ve stopped mentioning because they’ve assumed they are unsolvable?
• What alternatives and workarounds do they use today, and what does that tell us about what they actually value?
• What would need to be true—operationally, culturally, and technically—for our solution to be trusted and adopted at scale?

For the Axiom Health team, working through those questions revealed that the core problem was not inconvenience or cost. It was that existing exams produced qualitative, non-reproducible data that couldn’t add any predictive value, leaving physicians making consequential decisions about asymptomatic patients with instruments designed for acute diagnosis. The technology didn’t drive what came next. The friction did.

Pro-tip: Before your next product or strategy cycle, run a structured “friction audit” with customers and frontline staff. Surface the frustrations people have stopped voicing. Those shelved grievances often mark the most significant opportunities—and AI may have already made them solvable.

2. Redesign the Process, Not Just the Speed

“Lisa,” chief editor of a healthcare media startup, contacted us to collaborate on a strategic repositioning. The picture was stark: double-digit subscription declines, the loss of two anchor advertising partnerships, and a flagship content collaboration with a medical society that had quietly let its contract lapse. The board was asking hard questions. The team was exhausted.

Lisa’s first instinct had been to move faster: more content, tighter timelines, a leaner editorial process. When we began the engagement, a different diagnosis emerged. Her team’s publication, “Pulse,” wasn’t failing because it was slow. It was failing because it was building for a media model that no longer existed.

As research documents, digital publishing collapsed the advertising revenues that once sustained medical media, and open-access journals grew from 22 to nearly 17,000 by 2021, diluting any single voice’s authority. At the same time, the underlying market opportunity was growing. The healthcare digital content creation market is projected to reach $32 billion by 2030, driven by video, community, and personalized formats. The window wasn’t closing. It was shifting toward platforms and models that Pulse hadn’t yet built.

We helped Lisa and her team ask a different question: not how to produce more content faster, but what a media company genuinely designed around the modern clinician would look like. Drawing on models from industry news outlets, YouTube education channels, and AI-powered consumer newsletters, the framework they developed had three components.

1. Shift from content production to community curation. With AI handling synthesis, editors could instead focus on shaping stories and community. That freed up the team to launch a deferred YouTube channel, specialty newsletters by clinical area, and live case discussions that built a genuine practitioner community.

2. Use AI to personalize at scale. Rather than one newsletter for all physicians, AI-assisted segmentation allowed Pulse to tailor content format and cadence by career specialty, stage, and stated interest, creating the experience of a curated resource rather than a broadcast channel.

3. Look outside the category. Durable media businesses across industries had made the same pivot—from audience to community, from passive readership to active membership. Pulse had the clinical credibility to lead that shift in healthcare.

The leaders who win won’t be the ones who optimize their existing model. They’ll be the ones who redesign it around what their audience actually needs.

Pro-tip: Map your content’s “jobs to be done” for the practitioner: what decision, conversation, or action does your content help them take? If the answer is unclear, the format and cadence are probably wrong—regardless of how good the underlying journalism is.

3. Innovate in Micro-Pilots, Not Master Plans

Six months in, the board wanted a diversified revenue model: events, licensing, CME partnerships, sponsored research summaries. Lisa arrived at our next session with the question many overstretched leaders share: how do you innovate when you’re already running at capacity?

We pushed back. Recent research from the MIT Media Lab found that 95% of generative AI pilots delivered no measurable business impact, with most stalling because they were disconnected from real workflows (and lacked feedback loops that could help them learn and adapt). The answer to board pressure for innovation is not a bigger plan. It is a smarter experiment design.

What we proposed was a micro-pilot framework: structured, time-boxed experiments designed to generate learning quickly and fail cheaply. For each revenue stream, the team would spend no more than six weeks testing the smallest viable version, with AI accelerating both ideation and iteration.

Using AI to accelerate ideation velocity and stress-test format concepts (like sponsored series, CME modules, specialty subscriptions), the team surfaced ten viable ideas in the time it once took to develop a single brief. Lisa’s team launched three micro-pilots in the first cohort to pressure test pricing, licensing, and marketing: 

1. A specialty cardiology digest with a paid tier hit break-even subscriber numbers within five weeks. 
2. A sponsored research summary format generated strong advertiser interest but poor reader engagement and was modified. 
3. A marketing pilot was shelved entirely. The cost of the failure was two weeks of editorial time. The learning it produced shaped the next cohort of experiments.

The leadership posture the AI era demands is not the confidence to predict which innovations will succeed—it’s the discipline to find out quickly which ones won’t. AI compresses the time between hypothesis and evidence. Leaders who build that loop into their team’s work gain a structural advantage that compounds.

Pro-tip: For each strategic bet, define the smallest version you could test in thirty days with AI-accelerated ideation. Ask: what would we need to learn to make the next decision? That question, not the size of the pilot, is what separates useful experiments from expensive distractions.

The Imperative

The offsite conversation we opened with didn’t end with a bigger budget. It ended with a better question—and that question changed what the company built. 

The pattern holds across every engagement: the leaders who make the most meaningful progress are not those with the most resources or the most sophisticated tools. They are the ones willing to slow down long enough to ask what actually needs solving, to redesign instead of accelerate, and to treat a failed experiment as data rather than defeat.

Start with the friction. Redesign the process. Experiment at speed. These are not technology strategies. They are leadership disciplines. What has changed is that AI now makes them executable at a scale and pace that was previously out of reach for most organizations. The bottleneck was never the tools. It was the quality of the questions leaders were willing to ask.

Efficiency will keep the lights on. Imagination is what decides whose lights are worth watching.

View the full article