Sign in

A new category of business intelligence

Your business is complicated. Asking it a question shouldn't be.

Ask questions across the systems your business already uses. Synaptic Intelligence finds the relevant information, checks it, and gives you a clear answer — with the evidence behind it when you want it.

Question

How many employees do we have?

Answer

41. That's HR's count of people on an active employment contract today — and HR is the part of the business that owns what "employee" means. Payroll shows 43.

Why

The gap is real, not a mistake: two people were paid a final month after leaving, one long-term contractor stays on the payroll system, and one employee on unpaid leave hasn't dropped off it yet. Ask a tool that only sees payroll and you'd get 43, with no hint that contractors are being counted as employees.

Provenance

{{ f.kind }} {{ f.domain }}
{{ f.value }}
{{ f.note }}

A real answer from the product, shortened. Ask "why" and it shows this bridge — not just the final number.

One business. Different questions.

{{ hr.role }}

“{{ hr.q }}”

Your business is more than a collection of databases.

Question your business, not just its data. Most tools stop at the second one — here is what that costs you.

{{ m.eyebrow }}

{{ m.title }}

{{ m.body }}

{{ m.limit }}

What it means to treat a business as alive

Not a marketing image. Four properties of living things that a company shares, and that the architecture takes literally.

{{ l.eyebrow }}

{{ l.title }}

{{ l.body }}

Three kinds of number, never confused

{{ k.label }}

{{ k.headline }}

{{ k.body }}

A projection always names what it rests on.

Every important figure in an answer can be traced back to its source.

A live, verified answer in under 90 seconds.

Synaptic Intelligence is not retrieving a cached response or a good guess. It reads your live systems, checks each part of the business that might hold the answer, reconciles what comes back, and verifies every number before showing it to you. You watch it happen rather than waiting at a spinner.

{{ s.when }}
{{ s.stage }}
{{ s.body }}

Built to be deployed, not demonstrated

Three things a security review asks about on the first call

Who can see what, where the records actually live, and whether anyone can explain the shape of the thing afterwards. Each has its own page rather than a paragraph of reassurance.

Start with one question. Expand from there.

Synaptic Intelligence works with the systems your business already uses. You don't connect everything on day one — you connect what your first question needs, and widen it from there.

{{ ss.n }}

{{ ss.title }}

{{ ss.body }}

Questions people bring on the first call

{{ q.q }}

{{ q.a }}

Bring us one question your reporting can’t answer cleanly.

Think of a question that currently takes several systems, reports or people to answer. In forty minutes we’ll connect a read-only view of one system and show you what Synaptic Intelligence does with it.

Why this is different

Every other tool asks your data a question. This asks your business.

The distinction sounds semantic until you watch it fail. Data is what your business left behind: rows written at the moment something happened, by a system that had its own reasons for recording it that way. A business is the thing that produced those rows and is still producing them — with intent, with disagreements between its parts, with words that mean different things in different rooms.

Query the data and you get a technically correct answer to a question nobody actually asked. Question the business, and you get an answer that knows why the numbers look the way they do.

When you ask
Dashboard or AI chat over data
Synaptic Intelligence
{{ c.ask }}

Dashboard or AI chat over data

 {{ c.them }}

Synaptic Intelligence

 {{ c.us }}

Four things follow from the metaphor

Treating the business as a living entity is not a tagline sitting on top of a query engine. It is the reason the architecture has the shape it has, and each consequence is visible in the product.

{{ c.n }}

{{ c.title }}

{{ c.body }}

In the product{{ c.proof }}

Two objections worth taking seriously

Both are the first thing a technical buyer says, and both are fair. The answers below are the ones we give on the call.

{{ o.eyebrow }}

{{ o.q }}

{{ o.lead }}

{{ p.n }}

{{ p.title }}

{{ p.body }}

{{ o.limitsLabel }}

{{ l.title }}

{{ l.body }}

How we put it

{{ o.framing }}

What we are not claiming

A category claim invites scepticism, and it should. Three honest limits, stated before you find them yourself.

{{ n.title }}

{{ n.body }}

Entity visualisation

Your business, drawn as the thing it actually is.

An org chart shows reporting lines. A data catalogue shows tables. Neither shows the thing that decides whether an answer is right: which part of the business owns which word, what each one can actually read, and where two of them will disagree.

This is that picture, and it is live in the product rather than a diagram someone drew once: see what Synaptic Intelligence can currently reach, and what it can't yet. If a question needs a domain that isn't connected, the answer says so rather than quietly filling the gap.

Click any node. Then try a question and watch which parts of the business it wakes.

State {{ brainStateNote }}
{{ brainSvg }}
{{ bl.label }}

{{ bw.title }}

{{ bw.body }}

The graph is not decoration and it is not generated for the marketing site. It is drawn from the same registry the orchestrator routes against, which means it cannot show a connection the product does not have — and when a connector stops verifying, the picture goes hollow before anyone files a ticket.

How it works

One question in. One clear answer out — with the reasoning attached.

No single system knows your whole business, and one that claimed to would be lying. Synaptic Intelligence knows which part of the business to ask, keeps each specialist limited to its own systems, and combines what they say without smoothing over the places they disagree.

{{ st.n }}

{{ st.title }}

{{ st.body }}

{{ st.detail }}

Your systems don't always mean the same thing. Synaptic Intelligence knows the difference.

Finance counts an employee as a payroll subject; HR counts a signed contract. Sales counts a customer as an account; billing counts an invoice address. Neither side is wrong — they're simply counting differently, and a tool that quietly averages them is worse than useless.

Synaptic Intelligence keeps one definition per word — a contested term — in a business vocabulary shared across the company, and knows which domain owns each one. When that domain is connected, its answer is authoritative. When it is not, Synaptic Intelligence answers by proxy instead — another domain's closest measure — and says so permanently on the answer, naming what it used and how that differs from what you asked for. You don't have to settle these definitions yourself; Synaptic Intelligence already has.

{{ v.term }}
{{ v.owner }}
{{ v.definition }}

The vocabulary is yours to edit. Changing a definition changes every future answer that depends on it, and says so.

Pricing

You pay for the asking. Everyone else reads for free.

A monthly subscription for your company, with an allowance of questions in it. Answering a question does real work across your systems; reading one somebody already asked does not. So the price follows that line, and only that line.

Nobody is charged per head for looking at an answer. A regional manager who reads the weekly reconciliation costs nothing to serve, and charging for them would price a business out of sharing what it has learned.

What a credit is

One credit for each part of the business a question has to consult, and two more when several of them have to be reconciled into a single answer. Nothing else affects the price: not the length of the question, not how hard it is, not how much it reads.

{{ ce.q }}

{{ ce.credits }}

{{ ce.route }}

{{ ce.why }}

A question no connected system can answer is not charged for, and neither is a specialist that fails. You are billed for answers, not attempts.

{{ tr.name }}

{{ tr.badge }}

{{ tr.tagline }}

{{ tr.price }} {{ tr.period }}

{{ tr.credits }}

{{ tr.askers }}

{{ ti }}

{{ pn }}

On every tier, including the smallest

Nothing on this list is sold separately. A product that put sign-in, access control or provenance behind a higher price would be charging most for the thing every customer needs equally — and would contradict every other page on this site.

{{ ai.title }}

{{ ai.body }}

About the price

{{ p }}

Bring the question your reporting cannot answer cleanly.

Forty minutes, one read-only connection, your own question asked live. If a tier on this page is close but not right, say so — most of what we would change is the allowance, and that is a conversation rather than a rebuild.

Access & sovereignty

Everyone sees only what they are authorised to see.

Access is role-based and checked on every request. The intern and the CFO can ask Synaptic Intelligence the exact same question and reasonably get different answers back — not because the facts changed, but because each person is only shown what they are allowed to see. A single set of credentials for whoever wired it up, where everyone sees everything, is not a feature; it is the thing an enterprise buyer rejects a demo for.

The same shape holds up the org chart. A CEO's view spans the whole business; a divisional head's spans their division; a regional manager's spans their region; a functional executive sees what their function needs. One Synaptic Intelligence, with everyone seeing only their own authorised slice of it.

Access here is two separate things, kept separate on purpose.

{{ ax.eyebrow }}

{{ ax.title }}

{{ ax.body }}

Conflating the two is the mistake worth avoiding. A job title is exactly the sort of thing a person expects to edit about themselves — and if it also widened what they could read, editing your own profile would be a privilege escalation with a friendly form around it. So neither is self-service.

Who can do what

Four roles, and the operator role is not one your administrators can hand out.

Capability
{{ rc }}
{{ rr.what }}
{{ cell.mark }}

Asking is metered separately from reading because answering does real work and reading does not. A viewer can read every answer already produced without being able to spend a credit producing a new one — which is also how the works.

Six rules the model rests on

{{ ar.n }}

{{ ar.title }}

{{ ar.body }}

Data sovereignty

The safest copy of your data is the one nobody made.

Most analytics products begin by copying your records somewhere they can be queried quickly. That copy is then a thing to secure, to keep current, to delete on request and to explain to a regulator. We do not make one.

{{ sv.label }}

{{ sv.body }}

What is kept, and why

An audit trail is the one thing worth retaining, because a figure you cannot reconstruct is a figure you cannot defend six months later in front of someone who was not in the room.

{{ ap }}

If your security review has a question this page does not answer, ask it directly — we would rather answer it now than in a questionnaire later.

Trust and provenance

A number you cannot trace is a rumour with a decimal point.

Most systems that answer in natural language sound equally confident whether they read a record or invented one. The design of this product is a refusal of that. Confidence is expressed structurally — in what the interface shows and withholds — rather than in adjectives.

{{ ts.title }}

{{ ts.body }}

{{ it }}

What the product will not do

{{ r.title }}

{{ r.body }}

Security and data handling

The short version is below. Who may ask what, how tenants are separated and where records are allowed to live have their own page — .

{{ sec.label }}
{{ sec.body }}

Certification status and the current sub-processor list are available under NDA. Ask for the security pack.

Domains and connectors

Specialists, not modules.

Each of these is a domain: a specialist inside your business — finance, HR, sales — with its own connected systems, its own vocabulary, and its own read-only permissions. It answers only what it actually knows, and says so when a question falls outside that. Connect a domain and you can ask it anything within its area; disconnect one and the gap in what you can ask becomes visible, not hidden.

{{ d.name }}

{{ d.status }}

{{ d.description }}

{{ d.listLabel }}
{{ i }}

{{ d.note }}

Connecting a system

Connections are read-only by construction. A domain receives a scoped credential and a declared list of the tools it may call; anything outside that list is refused by the runtime rather than by good behaviour. Most connections are a database replica, a REST endpoint, or a scheduled export drop.

{{ cs.eyebrow }}

{{ cs.title }}

{{ cs.body }}

Questions

Asked often, answered plainly.

{{ p }}

Request a demo

Bring one question your reporting cannot answer cleanly.

Forty minutes. We connect a read-only view of one system, ask your question live, and walk the provenance of the answer with you. If the answer is thin, you will see exactly which missing connection made it thin.

{{ formNote }}

{{ cb.label }}

{{ cb.value }}

{{ cb.note }}
Good fit

{{ fl }}