Guide

How AI market intelligence works

An AI market intelligence system runs the same five stages every night: it collects documents from a defined set of sources, clusters near-duplicates so one event counts once, scores what is left against stated criteria, checks how many independent sources carry each surviving item, and writes the few that clear the bar into a brief with their sources attached. Most of the value is in stages two and three, where the volume is thrown away.

Stage one: collection

Each source is queried for the companies or topics being watched. This is ordinary engineering: fetch, parse, normalise into a common shape with a title, a link and the date the source itself gives.

The failure here is silent. A source changes its response shape or starts rate limiting, the connector returns nothing, and the night simply produces less. A system worth using reports per source what answered, what legitimately had nothing, and what failed, because those three look identical in the output otherwise.

Stage two: clustering

One event produces many documents. A funding round is covered by a dozen outlets within a day, and each copy looks like a separate finding to a naive system.

Clustering groups documents that describe the same event, usually by comparing overlapping word sequences rather than by exact matching, so a rewritten headline still lands in the right group. Done before scoring, it is what stops a quiet week reading as a busy one.

Stage three: scoring

Each cluster is scored so that only the top slice continues. The important design choice is whether the scoring is deterministic or a model judgement. Deterministic scoring - weighted trigger patterns, source authority, recency decay - is reproducible, explainable and cheap enough to run on everything.

This is also what makes the economics work. If ninety-nine per cent of collected documents never reach a language model, the model cost applies to the remaining one per cent, and the system can afford to read widely.

Stage four: corroboration

For the survivors, count how many independent sources carry the same event, by registrable domain rather than by outlet name, and discount copies that share wording with a press release.

The output of this stage should be a count and a list, not a verdict. Reported independently by three sources is a fact about the evidence. Verified is a claim about the world that no count can support.

Stage five: writing

Only now is a language model useful, and only on a small amount of material that has already been selected and bound to its sources. It writes what happened and what it means for the reader, and the link travels with the claim rather than being asked for afterwards.

The rule that keeps this honest: the model is given the documents and told to cite the ones it was given. It is never asked to remember a URL, because a remembered URL is a fabricated one.

Where QuikSignal fits

QuikSignal runs exactly these five stages nightly. One measured workspace night: 1,522 documents collected, 120 stories after clustering, 20 that a model was allowed to read, four that reached a person.

The other 1,518 are not deleted. They sit in a log in the workspace with the score that kept each one out.

See it on your own market →

What it does not do
  • Nothing in the chain predicts. It reports what was found and when.
  • The quality ceiling is the sources. A company that puts nothing out produces an empty brief, which the system says rather than filling.
  • A model is involved only at the last stage, on selected material. It is not asked to browse or to recall facts.
Questions

What people ask

Why not let a language model do the whole thing?
Cost and checkability. Reading every document with a model is expensive and slow, and a model asked to both find and judge tends to produce fluent output whose selection nobody can audit.
What does the clustering actually compare?
Usually overlapping sequences of words between documents, scoped to the same company and a time window. It is deliberately simple, because the goal is to catch rewrites of one story, not to understand them.
How would I know if a source had silently broken?
You would not, unless the system reports per-source outcomes for every run. That report is the first thing to ask for in a demo.

Read your own market the same way.

Eleven agents, the companies you choose, every night, with the document behind every line.