What is the Intelligence Gap?

Emulab VC12 July 20264 min read

Ask where investment conviction comes from, and most answers trace back to a surprisingly thin slice of the available evidence. The Intelligence Gap is the name for everything left on the table.

The signal that never arrives

By industry estimates, roughly 90% of decision-relevant signal lives in unstructured documents: patent filings, code repositories, regulatory registers, standards submissions, grant databases, technical papers. It is public, timestamped, and often startlingly specific about who is building what, and how fast.

Almost none of it reaches the models — or the people — meant to reason over it. Not because it is secret, but because it is voluminous, scattered, multilingual, and formatted for archival rather than analysis. A patent examiner can read one filing closely. Nobody reads forty thousand filings a quarter and holds the pattern in their head.

That distance — between the signal that exists and the signal that actually informs a decision — is the Intelligence Gap.

Why the gap decides who wins deals

Venture investing rewards being early and right. The industry's traditional answer to "how do you know?" is social: warm introductions, tier-1 co-investors, founder pedigree, conference heat. These signals are real, but they are also late by construction — they only fire once enough insiders already agree.

Technical signal fires earlier. A team's repository velocity changes months before its seed announcement. A patent family starts forming years before a product ships. A regulatory consultation forecasts a market's opening before any revenue exists. A fund that reads these sources systematically sees movement while it is still cheap to act on; a fund that waits for social confirmation pays the premium.

Why models alone don't close it

It is tempting to think a sufficiently large language model closes the gap by itself. It doesn't. Raw model access solves reading; it does not solve trust. An investment committee cannot act on a fluent paragraph of unknown provenance — and should not. Closing the gap for institutional capital means both halves: reaching the unstructured 90%, and delivering it in a form where every claim can be traced to its source.

DESIGN PROPERTY · Emulab is built so that a claim without a verifiable source is not made at all — grounding first, generation second.

What closing the gap looks like

Closing the Intelligence Gap is not a heroic one-time analysis. It is infrastructure: continuous reading of primary technical sources, disciplined extraction of comparable signals, and synthesis that shows its work. The output is not a hunch with better vocabulary — it is a cited case, auditable end-to-end, delivered while the window to act is still open.

That is the system Emulab is building for Southeast Asian deep-tech — the region where, as we argue in the companion piece on the SEA lag, the gap is widest and the reward for closing it is largest.

In short

What is it? The distance between available technical signal and the signal that actually reaches decisions.
Why does it persist? The signal is unstructured, voluminous, and scattered — impractical to read manually, untrustworthy when read naively by machines.
What closes it? Continuous machine reading, bound to citations, governed end-to-end — so speed never comes at the price of trust.