Fitted to your company

A market means what your sales team means by it.

PROWL is not a product you are handed. The vocabulary, the engine and the sources are adjusted to your business — because a chemical vocabulary describes what a substance does, and your company's vocabulary describes where the catalogue files it. Mapping between the two is knowledge about your company specifically, and it is the difference between a useful list and a list of things you already sell.

Three layers

One shared, one for the substance, and one that is yours.

Each company has its own way of saying and doing things, so the vocabulary is built in layers rather than as one dictionary. Your layer sits on top, and where it disagrees with the shared one, yours wins — and it is written down that yours replaced it.

Layer 1Shared

One shared map of market names, covering the industries any chemical might serve. Written once, used on every study.

Layer 2The substance itself

Who this chemical is: its search terms, and the false friends — the near-identical names that would otherwise pollute every result.

Layer 3Yours

How your company says things: your segment names matched to the shared market names, your own words for what you already sell, your brands and divisions, and the false friends that only bite in your context.

Why can't one vocabulary serve every company?
Because a company’s segment labels say where its catalogue files a product, and a chemical vocabulary says what the substance does. Map a producer’s own shelf labels onto a board of candidate markets and they may subtract almost nothing, not because the list is wrong but because the two vocabularies describe different things. One client’s shelf label is also meaningless to the next, which is why your layer lives in your own file and never in the shared one.
Which is why your words live in your own file. One company’s shelf label means nothing to the next.
Your systems

Plug in your own sources, and adjust the engine to your workflow.

The public record is the starting point, not the boundary. An engagement can bring in your own material — an internal product database, a document store, a subscription feed you already pay for — so that the exploration runs against what your company knows as well as what the world publishes.

What we set up for you

Your vocabulary, the list of what you already sell, your exclusions, your brands and divisions. These are settings. Changing one is quick, it does not touch how the engine works, and every entry records where the wording came from — so nothing is applied blind.

What we build for you

Connecting your own systems, and adjusting how the engine runs to fit the way your company works. That is real work done for your engagement, not a switch. It is also why the second chemistry is cheaper than the first — the fitting is done once.

What that is not, today. There is nothing you can plug in yourself today, and PROWL does not read your systems without you deciding what it may see. Saying otherwise would be an overclaim, and the honesty is the point — a finding you cannot check is a finding you should not act on, and that applies to claims about the engine as much as to claims about a market.

Is my data used to train an AI model?
No. Nothing here trains or fine-tunes a model on your data, and there is no such step in the system. What accumulates is the context the engine works from — your vocabulary, what you already sell, your exclusions and your evidence — held in your own files. Nothing you tell us is absorbed into model weights, which is why deleting your file genuinely deletes it.
What you keep is a library and a set-up of your own, not a share in somebody else’s model.
How it is governed

Every change is checked by a person, and written down.

A vocabulary anyone can quietly edit is worth nothing, because you can never tell whether a result changed for a good reason. So every change is proposed, reviewed against the evidence, and logged with a date.

Proposed

A change is suggested, with the evidence that prompted it attached.

Reviewed

Checked against that evidence by a person, not applied because a machine suggested it.

Logged

Dated, and recorded with whose say-so it was. You can always see what moved and when.

What is written down

Four things, each carrying its source and the date it was read.

The company and the market

The company — what you make and already sell, so served markets are subtracted before ranking.

The market — every candidate market and the facts behind it, each with its basis, its unit and its source.

The technology and the uses

The technology — properties as measured and published, not as claimed.

The uses — what the substance actually does, including how it behaves biologically, because that is often what decides where it can sell.

Two libraries, not one. What we learn about a market is public knowledge and carries no client name — it makes every study after yours faster. What you tell us lives in your own file and never crosses into the public one. The two are joined by a market reference and by nothing else. Delete your file and everything you gave us goes with it.

Why it compounds

The second chemistry is easier than the first.

The vocabulary, the record of what you already sell and the fitting all carry forward. Nothing has to be rebuilt to start the next product, which is what turns adjacent-market work from a project into a workflow you can sustain.

What compounds, and what does not?
The evidence, the vocabulary and the fitting compound. The judgement does not. Every gate still needs a person to decide it, and the last one still needs a signature — so the engine gets easier to run and sharper each cycle without ever becoming more able to choose.
A sourced library your company keeps. It does not depreciate.