YC coined the term “solution in search of a problem” to describe a project that starts with a cool solution but fails to find a real niche where it applies.

These projects often look fine on the surface, because their founders manage to duct-tape a half-fitting problem to the solution.

But anyone who’s worked on a SISP knows the feeling: you’re trying to convince others, and yourself, that the problem exists, when in reality it doesn’t.

I’m starting to get that feeling when I look at many AI-based software products on the market today.

Solution

LLMs are everywhere

Fundraising

Today, out of the last 10 funding rounds on startups.gallery, only one (Bluefish) didn’t involve an LLM-based product.

Most pitches I hear anecdotally now include LLMs as a central piece of the technology.

SaaS incumbents

Of the top 10 companies on the BVP Nasdaq Emerging Cloud Index, only two (Shopify and Adobe) haven’t added “AI” to their one-liners:

• Palantir: “AI-powered automation for every decision”
• Salesforce: “#1 AI CRM”
• Palo Alto Networks: “A new AI world is here”
• ServiceNow: “ServiceNow AI Platform”
• CrowdStrike: “AI-native cybersecurity”
• Snowflake: “AI data cloud”
• Workday: “The AI platform for people, money, and agents”

LLMs work for a few new companies

Some companies have created real value by embedding LLMs into their products:

• Cursor: I personally write more code thanks to Cursor.
• GitHub Copilot: in the early days of AI-powered IDEs, Copilot became popular by speeding up boilerplate coding.
• ChatGPT: it enables people to search and explore information in entirely new ways.

What about the others?

Do we see similar breakthroughs across the board in the Emerging Cloud Index?

Some companies had predictive technologies well before LLMs: Palantir (predictive law enforcement), Cloudflare (bot prevention).

Others have implemented tactical AI features, like Adobe.

But for most, I don’t see fundamental breakthroughs in their businesses driven by AI.

The MIT report

The MIT NANDA initiative’s The GenAI Divide: State of AI in Business 2025 claims that 95% of companies in its dataset have seen zero return on their AI pilots.

Many entrepreneurs I speak with laugh nervously when they hear this number. They’re not surprised: most organizations operate in outdated ways, and we’ve seen firsthand how these pilots fail.

Why is that?

Problem

AI is not your core business

Despite what the Emerging Cloud companies say, AI is not the core of their product.

For example:

• Notion is not an “AI workspace.” It’s a damn good note-taking system.
• Workday is not an “AI platform.” It’s an HRIS that isn’t horrible enough to make enterprises switch.

These companies reached their valuations not by slapping AI into their products, but by solving real customer problems.

Problems > Solutions

Anyone exposed to startup ideas knows that solutions are easy.

Most creative people can dream up elegant solutions to almost anything. Real problems are harder to find. The capitalist economy pays big money to those who solve problems, so a lot are already solved.

Jobs to be done

Truly spectacular product success doesn’t usually come from a new method, better ROI, or numerous features.

As Competing Against Luck notes with IKEA:

• IKEA operates in a saturated market, where it’s difficult to differentiate.
• It doesn’t win by selling “better furniture.” It wins by understanding the customer’s story: “I need to furnish my place today.”

That’s why rapid assembly and the Swedish restaurant matter. IKEA is the only furniture store where you can furnish an apartment in a day, and not starve while doing it.

IKEA stays a strong brand despite competing in a market most software founders would call abysmal.

Afterword

When evaluating a business or new idea, I’ve started to think away from AI.

Those two letters are irresistibly attractive to investors, so you autocomplete them into every pitch deck.

Maybe putting an LLM into everything really is the ultimate startup endgame. But there’s still a probability it’s just a trend, and delivering fundamental customer value will soon become fashionable again.