Business Intelligence Goldmine
Turning Connections into Opportunities.
A researched, fully cited view of your target market — where every single fact can be traced back to the page it came from, in one click.
Most market data cannot be checked.
Buy a list and you get rows of assertions. No source, no date, no way to tell what was researched from what was guessed. When a number turns out to be wrong — and some of them always are — you find out in front of a client.
The same weakness has become worse, not better, as AI has been pointed at the problem. A language model will produce a confident, plausible, completely invented company profile, and nothing about the output tells you which parts were real.
We built this the way we build an expert report: no assertion without a citation you can follow yourself.
Every fact carries its own evidence.
In this system a fact is not a value in a cell. It is a value bundled with everything needed to check it — and the platform will not store one without the rest. That is enforced in the database, so an uncited fact cannot be loaded even by mistake.
value- the fact itself
source_url- the live page it came from
archive_url- an archived copy, for when the live page changes
quote- the verbatim text that supports it
confidence- how strongly it is evidenced
as_of- the date it was true
Nothing is invented
Where the evidence is weak, the item is kept and marked as needing verification before you act on it — it is never quietly upgraded into a confident answer, and never deleted to make the numbers look cleaner.
Sources are archived
Web pages change and disappear. Alongside the live link we keep an archived copy, so the evidence behind a fact survives the site being redesigned six months later.
Facts are dated
Every fact records the date it was true. Refreshes update facts individually rather than replacing everything, so you can see what changed, when, and on what evidence.
An operated service, not a subscription to software.
You are not buying a tool and a login and being left to run it. We do the research, we verify it, and we hand you the finished view of your market.
- 01
Define the universe
We work out who actually belongs in your target market — by geography, sector, facility type, and the operational signals that matter to your offering. That definition is written down and agreed before any research starts.
- 02
We run the research
Our pipeline researches each company, pulls supporting evidence, and independently audits its own output. You do not run searches, buy credits, or manage tooling. This is the part you are paying for.
- 03
Verification gate
Researched data is reviewed and verified by a human in the loop to ensure accuracy and provenance. That control is built into the platform, so AI output never reaches you unchecked.
- 04
You work the data
Your team searches, filters, opens dossiers, follows citations, and exports. Refreshes run on an agreed cadence and apply as deltas, so you can see what changed and when.
What you get to work with.
Every company in your universe gets a structured profile — what they do, where they operate, the facilities they run, the equipment and service signals we can evidence. Facts are grouped by category, and each one opens its own evidence drawer.
A dossier is assembled from pages our tools actually fetched. The language model is never permitted to write down a citation of its own — it may only read what a retrieval tool returned and point at the exact sentence it relied on.
A separate, independent check then confirms that sentence really appears on that page. A quote that cannot be found is not published. That separation is the whole mechanism: the component that could invent a source is structurally denied the ability to record one.
Parent and subsidiary structures, shared ownership, and cross-links between companies inside your set. Where two targets are connected, the connection is clickable rather than something you have to already know.
One firm turns up across sources under half a dozen names — its legal name, its trading name, a subsidiary that owns a single plant, one spelling with “Inc.” and one without. Deciding when those are the same company is its own discipline.
Matching runs on official identifiers first and only then on judgement. Anything uncertain waits in a review queue rather than being merged automatically, because a wrong join is worse than a missed one: a missed join shows up as a visible duplicate, a wrong join silently contaminates every fact about both companies.
Named people in operational and buying roles, tagged by function and captured with the same citation discipline as everything else. Public professional information only — each record traceable to where it came from.
Contacts are captured with the same citation discipline as everything else, and restricted to public professional information — the role someone holds at a company, not their private life.
Where a record comes from your own CRM rather than the open web, it is labelled as such and kept private to you. First-party information is never dressed up to look like a publicly verifiable web fact, because the two carry very different weight when you act on them.
Where a client holds the appropriate data licence, project activity is linked to the companies in their universe — so a facility expansion or capital project shows up against the account it belongs to.
Project and facility activity is enriched using your own licensed data subscription, under your own key, scoped to your tenant alone.
There is no shared key and no pooled licensed dataset. What your subscription returns is stored as your private licensed data and is never visible to another client — that is both a security boundary and a licensing one.
We read your existing CRM, reconcile it against the researched universe, and show you what is missing, stale, or duplicated. Your CRM stays the system of record — we feed it, we never overwrite or delete it.
Your CRM is read, never written. We reconcile it against the researched universe and show you both directions: what your CRM knows that our research missed, and what we found that your CRM does not have.
Your own system stays the system of record. Often it is the better source — a real mobile number, the correct title, a relationship we would never surface from the open web. The point is to feed it, not to replace it.
Where sources contradict each other, both are presented side by side with their citations. The report shows you the conflict and its evidence; it does not label anyone or infer intent.
Where two sources disagree, both are presented side by side with their citations, their confidence, and the date each was true.
The report shows you the conflict and the evidence behind each side. It does not label a source as wrong, and it does not infer intent — that judgement is yours to make with the evidence in front of you.
Keyword and filtered search scoped strictly to your own data. Confident matches come back first; weaker candidates are returned separately and flagged to verify before outreach, never blended in.
Search is scoped strictly to your own tenant. Confident matches return first, ranked by match quality, then tier, then confidence, then recency.
When nothing confident matches, weaker candidates are returned in a separate tier and flagged to verify before outreach. They are never blended in with confirmed results, and an empty result is reported as empty rather than filled with something plausible.
A light working layer over the researched data — your own notes, statuses, and account fields, kept separate from the evidence layer so your commentary never gets confused with a cited fact.
A light working layer sits over the researched data — your own notes, statuses and account fields, kept in your tenant and never promoted into the shared research layer.
The separation is deliberate. Your commentary and a cited fact are different kinds of claim, and the moment they are stored in the same place, the citation guarantee stops meaning anything.
Your market intelligence stays yours.
Separation enforced by the database
Each client's data is walled off at the storage layer, not by application code remembering to filter correctly. A query that forgets the boundary returns nothing rather than someone else's market.
Licensed data stays with its licensee
Where a dataset is licensed to a specific client, it is tagged to that client and cannot be surfaced to another — including in aggregate views.
Access control and audit
Authenticated access with multi-factor sign-in, role-based permissions, and an audit trail of who looked at what and when.
Your notes never become our data
Anything your team writes into the working layer stays inside your tenant. Client-supplied information is never promoted into the shared research layer.
Built for teams selling into industrial and facility markets.
Sales and business development leaders
You know roughly who your market is but not precisely who is in it, who owns whom, or which facilities are actually worth a visit this quarter.
Operations and service organisations
You service equipment and facilities and need a defensible picture of the installed base and the operators behind it, not a scraped list.
Firms evaluating a market or an acquisition
You need a market map you can put in front of a board, where every figure has a source attached and disagreements between sources are shown rather than averaged away.
Where this currently stands
The research pipeline, the evidence and citation model, the client separation controls, and the CRM reconciliation tooling are built and in production use. The client-facing application is running as a working pilot and is being extended alongside the engagements it supports.
We are taking on a small number of new markets at a time so each one gets the research attention it needs. If your market is a fit, the first conversation is about defining the universe — not about software.
Our Clients Include:
Tell us what market you are trying to see clearly.
We will tell you honestly whether it is one we can research well, and what defining the universe would involve.