A synthetic retailer
One store started sending prices in cents.
A €26.99 item arrives as €2,699.00, in 1,703 rows, all from one store. If October's file reached the monthly report, sales would read €13.2M instead of €1.6M.
Watch the 90-second demoFor heads of data
Find the bad file before it reaches a figure someone signs.
One wrong export can move a month's numbers. Datainspec checks every delivery against the version you trusted, on your own machines, and leaves a record of every decision.
Two ordinary mistakes
What one bad file costs.
Both stories run on synthetic data built for our videos. The findings are the engine's own; the figures are our reading of the files.
A synthetic population register
One office opened its file in Excel and saved it.
84,134 person IDs turn into numbers like 7.42198E+12, and 3,685 IDs now belong to 7,501 people. A payment run matching on the ID would miss 17,282 people, €9.0M in a month.
Watch the 70-second storyWhy teams bring it in
Less firefighting, more evidence.
No project to set up
It learns every column from the version you approve. There is no rules file to write before the first scan.
How it worksYour data stays in-house
It runs on your machines and sends nothing out, no telemetry included. Built first for banks, insurers, healthcare and public bodies.
SecurityPeople decide, and it's recorded
Each finding goes to a person who marks it a real problem, an expected change or noise. The pipeline gets a verdict; the audit log keeps who, when and why.
GovernanceMeasured, not promised
585 of 628 faults planted in real public datasets caught (93.15%) with no rules written, and nothing raised in 432 scans of unchanged files. A slow drift that never trips an alarm still shows in History, column by column.
ResultsThe difference
From finding out late to finding out first.
Without it
With Datainspec
Errors found after the report went out
Found when the delivery arrives
Checks that depend on who remembers
Every delivery checked against the version you approved
Data sent to an outside tool to be checked
Runs on your machines; nothing is sent out
No record of why a file went through
Who decided, when and why, on record
How you'd start
A pilot, on one dataset that matters.
A few weeks, on your own infrastructure, with success criteria agreed in writing before it starts.
- 1Pick one datasetA recurring file or table that matters to you, and the version you trust.
- 2Agree what success meansIn writing, before anything is installed.
- 3Install on your sideOn your machines. Your data stays there.
- 4Review togetherWhat it caught, what was noise, and whether you keep it.
Questions
What leaders ask.
Does it replace our data quality tests?
No, it complements them. Your tests check what you thought of; Datainspec checks every column of every delivery against the version you approved, including what nobody wrote a test for.
How soon is it useful?
As soon as a version is approved: it profiles every column itself, so the first scan needs no rules.
Who sees our data?
Only the people with access to your machines. Nothing is sent to us: no data, no profile, no finding, and no telemetry.
What do you need from us for a pilot?
One recurring dataset with a version you trust, a machine to install it on, someone who knows the data for the review, and agreement on what success means before we start.
Find out what your last delivery hid.
One recurring dataset, your machines, a review at the end.