Why Most Transformations Fail and What Works Instead as discussed with Usman Sheikh on The Meg and Amy Show

"Your people aren't your problem, your model is."

That's the blunt diagnosis from Usman Sheikh, founder of High Output Ventures , after spending 18 months studying why professional services firms are struggling with AI transformation. His conclusion challenges everything most leaders believe about organizational change.

The problem isn't employee resistance to AI. It isn't a lack of technical capability. It isn't even insufficient investment in new tools.

The problem is that business models designed for human leverage are fundamentally incompatible with AI leverage. And most companies are trying to force-fit AI into systems that need to be rebuilt from first principles.

Sheikh's revelation came through working with a private equity fund doing accounting firm rollups. "I don't come from a professional service background," he explains. "This was my first foray in figuring out how these traditional businesses, which were built for a world of human leverage, were operating and how they were struggling with the shift to technology leverage."

What he discovered was both illuminating and disturbing.

These firms could clearly see the writing on the wall. Partners understood that AI would eventually transform their industry. They had the resources to invest in new technology. Many had already experimented with AI tools and seen promising results.

But they weren't transforming. Why?

The answer came down to game theory and incentive structures. As one leader told Sheikh: "I have no incentive to do this first. Because if I do this first and it doesn't work out, technology takes longer to catch up, I get whacked. And my compensation is not tied to that outcome directly."

The brutal reality: The market punishes short-term disruption to proven revenue streams, even when that disruption is necessary for long-term survival.

Sheikh's insight becomes even more powerful when you understand the organizational context we've been exploring on the show. As we discussed in our recent episode on organizational design, the communication complexity in traditional firms grows exponentially with size: a 100-person company has nearly 5,000 potential lines of communication.

But here's what Sheikh adds to that equation: When your business model is built on human leverage, adding more people theoretically adds more capability. When AI can handle specialized tasks, adding people without rethinking the model often just adds complexity without proportional value.

"Fundamentally, you're trying to augment people into workflows which need rethinking," Sheikh explains. "But at the core of it, the business model is the one which is causing the most friction. They're trying to force fit into the system. But if the plane is changing, then we may have to rethink this from first principles."

Sheikh's analysis reveals why startups have such a massive advantage in the AI economy. It's not just about being "more agile" or "less bureaucratic."

New companies can build with AI-first principles from day one. They can design their value creation, organizational structure, and operational processes around the reality that AI workers will handle certain tasks while humans focus on others.

Legacy companies, meanwhile, are "trying to retrofit AI onto business models designed for human leverage." They have existing revenue streams, established processes, and compensation structures that all reinforce the old model.

Sheikh points to compelling examples of this NewCo advantage:

Base44: Went from concept to $80 million acquisition by Wix in just 9 months. "That's not in the future. That's 2025. Someone sat there and thought, wait a minute, I can build an AI-native business to tackle problems that in the past would have required a whole team."

Harvey AI: Reached a $5 billion valuation in roughly three years by approaching legal services with AI-first principles, serving both general counsels and law firms with AI-enabled legal support.

These aren't outliers - they're early indicators of what becomes possible when you build for the new reality instead of retrofitting the old one.

Sheikh doesn't just diagnose the problem - he grapples with the fundamental question we explored on the show: Can legacy companies successfully transform into what might be called "RefreshCos" - legacy organizations that successfully reinvent themselves for the AI economy?

His framework is clear: there are "LegacyCos" (traditional professional service organizations with partners, billable hours, hierarchical structures) and "NewCos" (tech-enabled firms built on flywheels, ontologies, workflows, micro enterprises, and AI agents). But what about the companies in between - those attempting transformation?

Sheikh is skeptical about transformation at scale. "The challenge with RefreshCo is if the employee numbers become 50,000, 100,000, 500,000, is that possible? Refreshing at that scale is very difficult." He points to examples like Ford trying to do EVs inside the existing business: "It becomes very difficult to have two competing businesses run together with leadership."

But he acknowledges it's early days. "It is possible. It requires leadership, requires a whole bunch of pieces to put into place. Microsoft and some of these organizations are great examples... that have done exemplary well in certain parts of their business."

Two Paths Forward

Given the challenges of full-scale transformation, we discussed two more targeted approaches:

Sheikh's Spin-Out Strategy: "New co-spin outs with other external investors included in it. The main entity can hold 50%, but there has to be greater accountability from other counterparts." The key insight: it needs to be "cannibalizing parts of legacy co's businesses from the outside" with external accountability that prevents the parent company from imposing transformation-killing guardrails.

The Strategic Pocket Approach: As we discussed, legacy companies can identify areas where "people are heavily leveraging AI, are starting to use agents in repeatable ways, not just personal productivity improvements, but actually getting their whole team running better." These become proof points that can "show the rest of the organization what's possible," potentially creating momentum for broader transformation.

But as we discussed, there's a deeper challenge: we need to prove that new business models actually work at scale. This is where the historical perspective becomes crucial - every major business model shift required pioneers to demonstrate viability before widespread adoption became possible.

This insight about proving the model highlights why both approaches - spin-outs and strategic pockets - are essential. Sheikh's NewCo framework, whether implemented through spin-outs or strategic pockets, operates on three key layers:

1. Understand Your Value Creation Flywheel "If you don't get that right, nothing else works," Sheikh emphasizes. Most companies understand their current processes but not their fundamental value creation. When you understand where value actually comes from, you can adapt to any technological change.

For example, Sheikh shared the story of a consulting firm that saved a government $20 million on AWS deployments. The traditional firm would focus on the process: hours billed, people deployed, project management methodology. But this New Co understood their value flywheel: deliver measurable savings → get paid percentage of savings → reinvest in better AI tools and data analysis → identify bigger savings opportunities → attract larger clients seeking proven ROI → deliver even bigger savings. Because they understood this cycle, they could structure everything around outcomes rather than inputs: smaller teams, AI-enabled analysis, risk-sharing compensation models.

2. Rethink How Value Gets Delivered Traditional org structures break work into sales, marketing, support, product, finance. "This is where the deviation happens now," Sheikh says. "How does the value get delivered from the flywheel?" Instead of optimizing existing functions, successful refresh cos redesign how work flows based on AI capabilities.

Using the same consulting example: A traditional firm would have separate sales teams to find clients, consultants to analyze infrastructure, project managers to coordinate, and account managers to maintain relationships. But when your flywheel is savings → payment → better tools → bigger opportunities, you organize differently. This NewCo likely had cross-functional pods where AI analysts, client relationship managers, and outcome specialists work together on each engagement, all focused on maximizing and proving savings rather than optimizing departmental efficiency.

3. Design for Outcomes, Not Efficiency The winning companies Sheikh works with achieve "fewer people, higher leverage, delivered outcomes." But the key insight is that they're not just doing the same work more efficiently—they're delivering fundamentally different value propositions enabled by AI.

The consulting firm above exemplifies this perfectly. A traditional efficiency approach would be: "How can we do infrastructure audits faster using AI tools?" But the outcome approach asks: "How can we guarantee savings and get paid based on results?" This leads to completely different solutions - AI that can model complex cost scenarios, real-time monitoring systems, and risk-sharing business models that traditional firms could never offer. The result: $20 million in savings with a small team, versus traditional firms that might deploy dozens of consultants to deliver a report with uncertain ROI.

Sheikh's most sobering insight is about leadership. "Who cares enough to do these things? You have to really put yourself on the line to make some of these changes happen. Most things come down to leadership."

This connects directly to what we've been discussing about the need for leaders who can navigate uncertainty and redesign organizations in real time. The leaders who succeed in this transition won't just be good at execution—they'll be good at continuous adaptation.

But here's the catch: The compensation incentives for this kind of leadership don't exist yet.

"It comes down to compensation incentive structure," Sheikh notes. "Bezos comes to mind—when he was creating AWS, there was a period where he said, trust me, we're investing in this architecture and everyone hammered the stock, but you had to have faith in the leader."

This creates a critical gap. The market needs leaders who can navigate long-term transformation while managing short-term performance expectations. But the incentive structures punish exactly this kind of leadership.

However, this is changing faster than most people realize. As we explored in our conversation with Gustavo Alba, executive search leader and partner at Heidrick & Struggles , boards are already starting to evaluate leaders differently - looking for learning agility and business model innovation capability rather than just traditional P&L management. The leaders who develop these capabilities now, before the compensation structures fully catch up, will have enormous competitive advantage when the market inevitably shifts to reward transformation leadership.

Based on Sheikh's insights and our organizational design work, here are concrete steps leaders can take:

1. Start with Constraints, Not Technology Don't ask "How can AI make us more efficient?" Instead, set operational constraints: "Our mission is to onboard customers 20% faster" or "Handle a thousand more customers next year without increasing headcount." Frame challenges as outcomes to achieve rather than technology to deploy.

2. Map Your Value Creation Flywheel Sheikh recommends starting with Jim Collins-style flywheels. "Understand where value is created, because that is at the core of every business." If you only understand your current processes, you're trapped by them.

3. Create Guardrails for Experimentation Sheikh emphasizes that transformation requires empowering teams to experiment: "You need to create broader guardrails, delegate judgment down, and give greater autonomy." Without this, transformation becomes impossible because "if it's concentrated on the top and a certain group of people make all the decisions, coherence is going to be a challenge."

4. Find Your Wedge Sheikh sees the greatest success in "smart wedges into growing segments." Find areas where you can build AI-first capabilities without disrupting your entire revenue stream, then expand from there.

Sheikh's message is urgent but not alarmist. "Someone needs to start the conversation," he says. The companies that begin building AI-first capabilities now will have enormous competitive advantage. Those that wait for perfect clarity or aligned incentives will find themselves playing catch-up in a game where the rules keep changing.

The most important insight from Sheikh's work might be this: The technology isn't the hard part. The hard part is having the courage to cannibalize your existing business model before someone else does it for you.

As we continue to explore on the show, this isn't just about individual companies—it's about rebuilding the entire foundation of how organizations create value, structure work, and develop people in an AI-enabled economy.

The RefreshCo model offers a path forward. But it requires leaders who care enough to put themselves on the line for long-term success over short-term safety.

The question isn't whether business models will need to transform—it's whether you'll lead that transformation or be transformed by it.

Prepared by Amy Wilson, former tech executive and current product strategy advisor. For more insights on leadership and the future of work, subscribe to The Meg and Amy Show.