How much of your market can AI actually see?
Findings · Jul 2026 · 4 min
Every founder is told to stay close to the market. Almost nobody says how much of it your tools can actually reach. So we ran the test. Five AI research tools, the same topics, one honest count of what each found. Most saw under 15% of the available sources. Aim saw 89%. Here's how we measured it, and why the gap should bother you.
How we ran the test
The method is plain on purpose. Pick a topic. Ask five tools to surface everything they can find on it. Pool every source they return into a single list. That list is the denominator: the union of everything any tool found.
Then score each tool by the share of that pool it actually surfaced. No private list. No home advantage for anyone. A source only counts if at least one of the five could reach it.
That last rule is generous to the field. Anything none of the tools found never enters the pool, so the real market is bigger than what we measured. These are best-case numbers for everyone in the test.
What we found
ChatGPT deep research surfaced 8.5% of the pool. Claude with search, 10.2%. Exa, 11.3%. Perplexity, 14.4%. Aim surfaced 89%.
The best of the mainstream tools stopped under 15%. Aim cleared 89%. That gap isn't a rounding error or a few clever sources. It's most of the market, found by one tool and missed by the rest.
Everyone's wrong together
Under 15% has a specific meaning. You're deciding on the first page of results. The loud, obvious, already-indexed surface every tool reaches first.
Because every tool grabs the same surface, everyone misses the same slice. Your competitor's AI is as blind as yours. The consensus feels solid because it's shared. Shared isn't complete. 'Everyone knows' usually means everyone knows the same 15%.
Under 15% isn't a smaller view of the market. It's the first page of results, mistaken for the whole thing.
Where the other 85% hides
The missed slice isn't scattered noise. It clusters in the formats crawlers struggle to read: podcasts, video, keynote and earnings-call transcripts, founder posts, niche trade press.
That's where people say what they actually think, before it's sanded into a press release. From podcasts and video alone, Aim surfaces roughly six times more facts than every other tool in the test combined.
Six times. From the formats everyone else skips.
What changes at 89%
Coverage isn't a vanity metric. It's the difference between reacting to what surfaced and seeing what moved.
At 89%, you stop guessing from scraps. You catch the shift when a founder floats it on a podcast, not a quarter later when it's a headline. Your agents pull cited context on the market as it's actually discussed, not the market the first page describes.
Staying close to the market was always the advice. Coverage is how you finally do it.