Business intelligence, answered
Say it. Get the fact.
Ask a question about your business in plain language and get a trusted answer back. No analyst in the middle. No dashboard to build first.
Not ready to talk yet? Watch a minute in the product, or read the questions it answers.
Built for the people who run the business
If you run distribution, multi-store retail, manufacturing, or a franchise network -- and you do not have a dedicated analytics team -- this is for you. Large enterprises with specialist BI teams are not the first customer.
Ask in plain language. Get an answer you can trust. Some numbers are a standard formula this warehouse can already compute. An administrator vouches for everything else. It runs on your premises, on your warehouse, for your company only.
The short version, and a minute in the product, is on What is FactSmith.
Questions like these
The product starts from the question you have now -- not a dashboard someone designed six months ago.
Distributors ask which customers are buying less, where margin is disappearing, and which stock is slow-moving.
Multi-store retailers ask which stores are underperforming and why, what changed after a promotion, and how store profitability compares.
Manufacturers ask which products are most profitable, where capacity is being lost, and how demand compares with production.
Franchise groups ask which sites need attention, how margins differ by franchisee, and what changed by region.
Every one of those is answered in full on Business questions.
Same product for each. Different questions. Sector pages: Distributors, Retail, Manufacturing, Franchise. Path: How it works.
What a first session looks like
Connect your warehouse. Agree what revenue, customer and on time mean once. Ask. Keep the answers worth keeping.
There is no modelling project and no BI rollout. An evaluation copy is a folder and an installer. It works before you give it an AI key.
Four steps on How it works. A minute on screen on What is FactSmith.
What is here today
Available now
FactSmith Ask
Answers out. Available now.
Early development
FactSmith Easy Data
Data in. Being built now.
On the roadmap
The wider platform
Cloud identity, Teams and Copilot, an aggregate engine that learns from use.
Returning often? What is new lists recent changes on this site. Every clip is on Videos.
The people with the answers are drowning in the work
Rebuilding the same report every week. Stitching spreadsheets together. Reconciling two numbers that should have matched. Explaining results instead of investigating them.
The people who know the business best spend their week producing numbers rather than acting on them.
The problem was never the questions
Why are margins falling? Which customers are actually profitable? What changed this month? Every business has these. The cost is everything that happens between asking and knowing.
Answers, not dashboards
A dashboard answers the question someone had six months ago. FactSmith answers the one you have now, and shows you what it did to get there.
Ask it how you'd say it
Speak the question. FactSmith transcribes it and shows you the text before it runs, so you can correct a misheard product name instead of wondering why the answer looks wrong.
Where data becomes fact
Between your systems and your answer sits a layer that knows what your business means by revenue, customer and on time. That is what makes an answer trustworthy rather than merely fast.
Every answer explains itself
Ask any answer why it looks the way it does and it tells you: which fields it used, and in plain language why it drew this chart rather than another. You can read the query it ran, and refine it by asking again.
Keep the answers that matter
Save any answer, and it is there tomorrow. Group the ones you check regularly into a dashboard that persists -- yours, named, laid out how you want it.
Ask your dashboard what is happening
Turn a dashboard into a written brief. FactSmith reads the tiles on it and produces an executive summary of what is happening and why, with the charts embedded and the whole thing exportable to PDF. The numbers are calculated, not narrated into existence.
Someone has to say it is true
Some answers are a standard formula this warehouse can already compute. Those are facts without a person signing them, and they stop being offered if the inputs go away. Anyone can save any other answer. Only an administrator can vouch for one of those.
Without that second step every saved number drifts towards looking official, and the word trusted stops carrying information. A vouched-for fact teaches the system how your business asks questions. It never answers on its own, because a fact confirmed in error would then be repeated identically forever, and nobody would notice.
Know when to pay attention
Set a target for a measure, without waiting for a release. Tiles colour themselves against the target behind their numbers, so a dashboard shows you what is off track before you have to read it.
Know what it costs, and where it goes
Choose which AI model answers your questions. Before you choose, FactSmith shows what each one costs per question and states plainly whether using it sends anything outside your network.
Every question is recorded with what it cost to produce. Including the ones your permissions refused: those cost real money too, and leaving them out would under-report the bill.
Why you can trust the answer
Permissions are enforced beneath the AI, not by it. A regional manager asking about revenue gets their region, because the rest of the data never reaches the model.
It runs where your data already lives
One company, one source database, one deployment. Not a shared platform holding everybody's numbers in the same place.
An evaluation copy is a folder and an installer. It needs Docker Desktop and nothing else, and no internet connection once it has reached you. It works before you give it an AI key: the questions it suggests have verified answers stored with them, so a first session works end to end out of the box.
Every business deserves access to its own intelligence
See FactSmith answer questions about your own data.