The AI Problem Matrix: Where the Largest Opportunity Hides

Not all AI problems are created equal. Some run out of work. Others never do. The difference determines the size of the opportunity.

I’ve been mapping AI businesses across two axes. The first: open loop versus closed loop. Closed loop systems validate their outputs. An AI coding agent runs tests, checks if the code compiles, confirms the feature works. Open loop systems generate without validation. An AI writing marketing copy can’t prove the copy sells product. The second axis: finite versus infinite demand. Finite demand exhausts. A customer support team handles a fixed number of tickets per day. Infinite demand never ends. There’s always more software to build, more content to create.

The combination creates four quadrants. Closed loop plus infinite demand: coding. Every company needs more software. The backlog never empties. GitHub Copilot, Cursor, Devin all chase this market. They solve real problems with measurable outcomes. Tests pass or fail. Features ship or don’t. The feedback loop is tight.

Closed loop plus finite demand: customer support. Zendesk handles 10,000 tickets per day, not a million. The work caps at the customer base. AI assistants reduce cost per ticket, but they can’t expand the market size. The opportunity is bounded by demand, not technology.

Open loop plus finite demand: campaign design, presentation creation, one-off creative work. Marketing teams need a finite number of campaigns per quarter. The AI generates, humans approve, the work completes. Good business, limited scale.

Then there’s the fourth quadrant. Open loop plus infinite demand. This is where the largest opportunity hides. What work never runs out & requires creative generation? Content creation. Research. Ideation. Every company needs more blog posts, more social media, more thought leadership, more market analysis, more competitive intelligence. The demand is insatiable. The validation is loose: does this resonate? Did it drive engagement? The metrics exist but lag by weeks or months.

Why does this quadrant matter most? Three reasons. First, TAM. The market for content, research, & creative work spans every industry & every company. Second, defensibility. Open loop systems require judgment, taste, & domain expertise. They’re harder to commoditize than closed loop tools. Third, margin. Creative work commands premium pricing when it works. A great marketing campaign outperforms a good one by 10x, not 10%.

The startups building here will look different. They won’t measure success in tickets closed or tests passed. They’ll measure in engagement, reach, & influence. The feedback loops are slower. The validation is fuzzier. But the demand is unlimited. That’s the quadrant I’m watching. That’s where the next wave of $10B+ AI companies will emerge.

The pattern is clear. Closed loop captured the first wave: coding assistants, customer support bots, data analysis tools. Open loop with infinite demand will capture the second. The work that never ends, generated at scale, with human judgment in the loop. That’s the future.