This is the third post in the series “Problem First: AI-Assisted Problem Solving for Organizations That Can’t Afford to Get It Wrong.”

The first two posts in this series made a general case. Organizations are bad at defining problems. They reach for solutions before understanding what’s actually wrong. And the structured process that could prevent this, the five-stage framework from framing through root cause analysis to action, is rarely practiced with any rigor.

Everything I’ve described so far applies to organizations of all sizes. Large corporations misdiagnose problems all the time. But large corporations have something that absorbs the cost of those mistakes: scale. They have budgets that can tolerate a failed initiative. They have teams that can regroup and try again. They have institutional memory, however imperfect, that accumulates lessons from prior failures.

Small and mid-sized enterprises and nonprofits have none of these cushions. And that’s what makes the problem-solving deficit not just inconvenient for them, but existential.

The Double Bind

The situation facing SMEs and nonprofits is best understood as a double bind.

On one side, these organizations have severely limited resources for implementing solutions—any solutions. A mid-sized manufacturer doesn’t have the capital to run parallel experiments. A community nonprofit can’t afford to staff two competing programs to see which works. Every dollar, every hour of staff time, every ounce of organizational energy is committed. There is no slack in the system.

This means that when an SME or nonprofit implements the wrong solution—one that addresses a symptom rather than a root cause, or that was designed for the wrong problem entirely—the consequences are disproportionately severe. A large corporation writes off a failed pilot and moves on. A small organization may not recover. The resources spent on the wrong answer are resources that can’t be spent on the right one.

On the other side of the bind, the very resource constraints that make getting it right so critical also make it nearly impossible to invest in getting it right. Structured problem solving takes time, attention, and often external expertise. SMEs and nonprofits lack dedicated innovation departments. They lack strategy teams. They lack the budget to hire consultants who might guide a rigorous diagnostic process. Their problem solving is, as I’ve described it elsewhere, ad hoc at best.

The double bind, then, is this: the organizations that can least afford to solve the wrong problem are also the organizations least equipped to define the right one. Complication over complication.

The Nonprofit Reality

This double bind is acute across the entire SME and nonprofit world, but it hits nonprofits with particular force—for reasons that are both structural and cultural.

Start with the structural reality. According to the National Council of Nonprofits, 88% of nonprofits in the United States operate on annual budgets of less than $500,000; 92% operate on less than $1 million. Only 3-5% have budgets exceeding $5 million. When people think of the nonprofit sector, they often picture large institutions: major universities, hospital systems, and well-known national charities. But the sector is characterized by what the Council calls a “long tail”: a small number of very large organizations account for most of the revenue, while the overwhelming majority are small, community-based operations serving local neighborhoods and specific populations.

These small and very small organizations are the core of the sector in terms of sheer numbers. And they are precisely the ones operating with the thinnest staff, the tightest budgets, and the least capacity for strategic planning of any kind—let alone structured problem solving.

Now add the cultural dimension. Nonprofits operate under mission-driven pressure that creates a unique obstacle to rigorous problem definition. In a corporation, questioning the framing of a problem is uncomfortable but conceptually acceptable—it’s a business decision. In a nonprofit, questioning the framing of a problem can feel like questioning the mission itself.

Consider a homelessness-focused nonprofit that has defined its problem as “insufficient shelter capacity.” Questioning that framing—asking whether the root cause might be something other than shelter shortage, perhaps failures in mental health services, housing policy, or employment support—can feel like an act of betrayal. The organization’s identity, its fundraising narrative, and its staff’s emotional commitment are often bound up in a particular understanding of the problem. Redefining the problem threatens all of that.

The result is that the organizations serving the most vulnerable populations are often the most resistant to the kind of rigorous problem interrogation that would make their work more effective. Not because they lack compassion or dedication—but because the structure of their world makes honest diagnosis feel dangerous.

The Knowledge Paradox

There is a further complication, and it’s one I’ve been thinking about since I first wrote about AI as “the great equalizer” for smaller organizations.

The promise was straightforward: AI tools, particularly large language models, dramatically reduce the cost of domain expertise. SMEs and nonprofits can now access sophisticated analysis, strategic frameworks, market intelligence, and scenario planning that were previously available only to organizations with dedicated strategy teams. The playing field, it seemed, was finally leveling.

But there’s a paradox hidden in that promise. The knowledge that AI makes abundant and cheap is explicit knowledge—information that can be codified, documented, and transferred. Reports, analyses, frameworks, data syntheses. This is exactly what AI excels at generating. And precisely because AI makes this kind of knowledge universally available, it becomes a commodity. If your AI can produce sophisticated market analysis, so can your competitors.’Explicit knowledge, once a source of advantage, becomes a common baseline.

What remains as a differentiator is tacit knowledge, the kind embedded in experience, institutional memory, professional judgment, and organizational culture. It’s the interpretive layer that determines which AI-generated insights matter and which don’t. It’s the accumulated wisdom that knows when to act boldly and when to proceed with caution. It’s the pattern recognition that comes from years of navigating specific markets, communities, or populations.

Large organizations have this layer, however imperfectly, built up over decades. They have experienced professionals who’ve weathered multiple cycles, institutional processes refined through trial and error, and networks of expertise that span functions.

Most SMEs and nonprofits don’t. They’re gaining access to an ocean of explicit knowledge while lacking the interpretive infrastructure to use it well. They’re rich in information and poor in wisdom. And the gap between the two is precisely where problem-solving failures live.

The Risk of Automating Dysfunction

All of this converges on a risk that I find genuinely alarming: that SMEs and nonprofits will adopt AI tools enthusiastically, integrate them into existing workflows, and end up automating their dysfunction rather than solving it.

The pattern is already visible. Organizations acquire AI capabilities and immediately ask: “Which of our processes can AI improve?” The question sounds sensible. It’s the same technology-centric trap I described in the first post—but with higher stakes, because these organizations have no margin for error.

A small manufacturer deploys AI to optimize a supply chain that was designed for a market that no longer exists. A nonprofit uses AI to generate more grant proposals without questioning whether its programs are addressing the right problem. A community organization automates its outreach without asking whether the people it’s reaching are the ones who most need its services. In each case, the AI works. The process it’s applied to doesn’t.

The adoption window for AI is compressing rapidly. We are likely to have only a few years until peak adoption. SMEs and nonprofits that spend this time celebrating access to AI-generated knowledge without building the problem-solving discipline to use it wisely will find themselves in a peculiar position: technologically current and strategically adrift. They’ll look modern. They’ll perform no better.

What’s at Stake

For SMEs, what’s at stake is competitive survival. In a world where explicit knowledge is universally accessible, the organizations that thrive will be those that can define their problems with precision, diagnose root causes rather than treat symptoms, and deploy resources, including AI, against the right targets. The agility that has always been the small organization’s advantage only matters if it’s pointed in the right direction. Moving fast toward the wrong destination is worse than moving slowly toward the right one.

For nonprofits, the stakes are different and arguably higher. When a nonprofit misdiagnoses the problem it exists to solve, the consequences extend beyond the organization itself. The communities it serves don’t get the help they need. The donors who fund it don’t get the impact they were promised. And the broader public loses confidence in the sector’s ability to address society’s most pressing challenges.

Consider the arithmetic. If 92% of nonprofits operate on budgets under a million dollars, and if even a fraction of those organizations is solving the wrong problems because they lack the capacity for rigorous problem definition, the aggregate waste—in money, in staff time, in missed impact—is staggering. Not because these organizations are careless. Because the system they operate in gives them no tools, no training, and no incentive to pause and ask whether the problem they’ve defined is the problem they need to address.

A Different Kind of Equalizer

I began this series by observing that organizations are consistently bad at solving problems. In the second post, I laid out what a rigorous process looks like—five stages, each with a clear purpose and a predictable failure mode when skipped. In this post, I’ve argued that the organizations most in need of that process are the ones least likely to have it.

This is where AI re-enters the picture—but not in the way most people expect.

The real value of AI for SMEs and nonprofits isn’t in generating answers. It’s in improving the quality of questions. When used within a structured problem-solving process—not as a replacement for one—AI becomes the thinking partner that these organizations have never been able to afford. It can help surface assumptions, challenge problem framings, map root causes, and stress-test solutions before resources are committed.

That’s the subject of the final post in this series: how AI tools, deployed with discipline, can begin to close the problem-solving gap—not by giving small organizations the same answers as large ones, but by giving them the ability to ask better questions.

Next in the series: “AI as Problem-Solving Partner: Doing It Right.”