PROWL · a new AI business-development engine

We translate a chemical into what it actually does — then find every industry in the world that needs that.

PROWL — Pattern Recognition Of White-space Leads — does not search your molecule's name. It works out the substance's functional language, reads the world's most difficult commercial and technical evidence for anywhere that function is already wanted, and tests each candidate for whether it is real, worth pursuing and winnable. A person makes every judgement.

The translation

A chemical's name tells you almost nothing about where it can sell.

This is the step that is usually missing, and it is the one that decides everything after it. Searching a trade name or a CAS number returns the markets you already serve, because those are the people who already use that word. The markets you do not serve do not call it what you call it. So the first job is to stop working in names and start working in function.

Step 1Identity

What the substance actually is, resolved against the registers: the identifiers, the chemical family, the forms it is genuinely sold in — and which of those forms have quietly ceased.

Step 2Function

What it does. Interfacial behaviour, thermal and oxidative stability, biological activity, compatibility. Properties as measured and published, never as claimed. This is the language industries actually buy in.

Step 3Industrial use, worldwide

Who needs that function, in any industry and any region — not one sector, and not one continent. The same behaviour is called different things in adhesives, in agriculture, in energy and in personal care, and each of those names has to be recognised as the same thing.

Step 4Market, in your words

Uses are grouped into markets through your own ontology, so what comes back is expressed the way your business describes itself — and what you already serve is subtracted before anything is ranked.

Why translate a chemical into function rather than search its name?
Because the industry that needs your molecule is, by definition, not the one that already buys it — and it does not use your name for it. Searching the name finds the markets you already serve. Searching the function finds the ones you do not.
This is the step that makes the rest possible, and the one a keyword search cannot perform at all.
The material

Most of the evidence is hard to read on purpose.

This is not a matter of volume alone. The documents that decide the question are written in four different professional registers, none of them written for a commercial reader, and each has to be understood on its own terms before anything can be compared across them.

Legal and regulatory

Registration dossiers, consent orders, restriction notices, food-contact inventories. Written by lawyers for regulators. Precise, deliberately narrow, and full of consequences that are never stated as conclusions — a declared use, a ceased identity and a missing entry each mean something quite different.

Patents

Drafted to claim as broadly as possible while disclosing as little as the law allows. Your chemistry usually appears as somebody else's input, unnamed, described by function or by family. Filings show intent rather than revenue, and are read that way.

Academic literature

Written for specialists in a field that is probably not yours, with the commercially decisive result sitting in a methods section or a supplementary table rather than the abstract.

Trade and market data

Customs and volume data carry numbers without narrative; research houses carry narrative without a register behind it. Each is recorded for what it is — verified in a register, or asserted by someone selling a report — and the two are never blended into one confident sentence.

Why does the difficulty of the material matter commercially?
Because the evidence that changes a decision is rarely in the document that looks promising. It is a clause in a dossier, a dependent claim in a patent, or a line in a supplementary table. Material like that is not skimmable, which is exactly why it goes unread — and why it is still there to be found.
Eight kinds of public source, read in full rather than sampled. Every fact carries its source and the date it was read.

The point is the range. Spanning the globe, across every industry rather than one — because the sector that needs your molecule is by definition not the sector you already sell to, it is probably not on your continent, and nobody in your building is subscribed to its journals.

Stage 1

Discover — find the opportunities

Public sources. The AI reads and translates; a person judges.

1Your chemical

Its identity, and what you sell today.

2Independent sources

Use data, regulators, patents, literature, trade data and the open web — eight kinds of public source.

3Evidence, not summaries

Each use becomes a record: the sentence, the document, a live link.

4Your markets subtracted

Grouped in your language. What you already serve is set aside and never shown back to you as a discovery.

What comes out of the Discover stage?
A ranked list of genuinely new markets — ranked by weight of evidence, not by guesswork. Every search is kept, so the list can be audited rather than believed.
Uses are grouped into markets through a custom ontology written in your company’s own business vocabulary.
Stage 2

Qualify — build the business case

Three questions, answered from named public sources — open sources before paid.

Is it real?Does it exist at scale?

And do independent sources agree that it does?

Is it worth it?As-is, or with work?

Does it sell as the molecule stands, or does it need formulation work first? Those are two different sales.

Can we win?The route in.

Who buys, who is there already, the route in, and the demand driver behind it.

A ranked list is not a plan.

What comes out of Qualify is not a score. It is the beginning of a commercial strategy for expansion into that market.

The sale, named

Does it sell as the molecule stands, or does it need formulation work first? Those are two different sales, to two different buyers, on two different timescales.

The route in

Direct to the formulator, or through specialty distribution. Who is incumbent, what they hold, and where the window is — measured in filings per year rather than in sentiment.

What is blocked, and why

Blocked markets are reported too, with what opening them would take. A closed door with a named lock is more useful than an open door that isn't there.

Who signs off on a conclusion?
A person, always. The AI drafts each conclusion and a person reviews and approves it before it goes anywhere. Blocked markets are reported too, with what opening them would take.
Every fact carries its source and date. The machine drafts; a person signs.
The knowledge store

The study ends. The library doesn't.

Most market work ends as a slide deck, and what was learned getting there is lost. PROWL ends as a structured library your company keeps — written in your own business vocabulary, every entry carrying its source and the date it was read. That library is what makes the second chemistry easier than the first.

The ontology

Your vocabulary, governed and versioned

How uses are grouped into markets your way, why your served business must be subtractable, and how changes are proposed, reviewed and logged.

 Accelerated and sustained

Turning this into a cadence, not a project

Why adjacent growth is normally done once and then never again, and what it does to the P&L when it runs continuously.
Market watch

A market you looked at once is a market you looked at once.

Evidence goes stale. A lead that was real in March can be taken, priced out or regulated away by September, and nothing about the original report will tell you. So PROWL keeps walking — on a cadence, against the same sources, reporting only what actually moved.

The cadenceLive leads every 30 days

Leads still in play are re-walked monthly; everything else quarterly. A market you already sell into is never re-walked — news cannot overturn the fact that you are already there.

The signalMovement, not rewording

Each walk is compared against the last on the evidence itself — the quoted source text, not our summary of it. A fact described in different words is not a change, and is not reported.

The honestyWhat we could not check

Every summary names the sources that could not be read this time. “We looked and found nothing” and “we could not look” are different facts, and you get told which one you have.

What does PROWL not do?
PROWL does not size markets, and it does not rank commercial attractiveness. A high score means a market is well supported by evidence — not that it is attractive. That judgement is made by a chemist with 30 years in the industry, working with you.
So a high score tells you where to look. It does not tell you what to do.