Demo
From a sheet nobody validated to a dashboard that answers questions
This company does not exist and this data is invented — which is why it can be shown. Everything else is real: the audit counts the errors one by one, the dashboard is computed from those rows, and the agent runs the query you will see on screen. Nothing is typed by hand.
Act 1
This is how the data arrives
Ninety-nine order rows, straight out of the spreadsheet that runs the operation. At a glance it looks fine. Press the button.
| # | Cliente | Ciudad | Valor | Teléfono | Correo | Vigencia |
|---|---|---|---|---|---|---|
| P-1000 | Comercial Bermúdez | Manizales | 2.989.000 | 3622849205 | comercialb@correo.com | 05-01 → 07-28 |
| P-1001 | Distribuciones Andina | MANIZALES | 2523000 | 3113811532 | distribuci@correo.com | 06-01 → 08-28 |
| P-1002 | Depósito San Martín | Manizales | 3727000 | 3546819870 3466167506 | depsitosan@correo.com | 06-01 → 08-28 |
| P-1003 | Herrajes del Norte | Manizales | 989000 | 3439874672 | herrajesde@correo | 08-01 → 10-28 |
| P-1004 | Ferretería El Roble | Cali | 2505000 | 3148644545 | ferreterae@correo.com | 05-01 → 07-28 |
| P-1005 | Depósito San Martín | Cali | $ 1.744.000 | 3122666068 | depsitosan@correo.com | 04-01 → 03-28 |
| P-1006 | Herrajes del Norte | Bogotá | 1.521.000 | 3010457337 | herrajesde.correo.com | 04-01 → 06-28 |
| P-1007 | Pinturas Océano | Bogotá | 2801000 | 3587579656 3439793111 | pinturasoc@correo.com | 03-01 → 05-28 |
| P-1008 | Importadora Zafiro | CALI | 3634000 | 3126379083 | importador@correo.com | 08-01 → 10-28 |
99 rows
109errors across 71 rows
- 20phone numbers stuck togetherTwo numbers in one cell. Nobody can dial that, and on export the destination system truncates it without a word.
- 24emails that are not emailsMissing the domain or the @. They look fine at a glance and bounce every time.
- 20cities spelled differentlyThe same city in caps or without its accent. Each variant is a row that drops out of the total with nothing to warn you.
- 33amounts stored as textWith separators, with a currency symbol. Summing that column gives you a number, and it is the wrong number.
- 9impossible date rangesThey end before they begin. The spreadsheet accepts them without complaint.
- 3duplicate ordersThe same order counted twice. The most expensive one, because it inflates sales and nobody questions it.
None of these raise a warning. The spreadsheet stores them just as happily, and Monday’s report is built on top.
The database Cauce builds rejects all six in its own structure. They do not depend on somebody noticing.
Act 2
This is how it ends up
The same rows, without the duplicates and with the cities unified. The dashboard is not assembled by hand: every figure is computed from those rows as the page loads.
- MedellínCOP 61,088,000
- ManizalesCOP 48,505,000
- BogotáCOP 42,003,000
- BarranquillaCOP 39,370,000
- CaliCOP 38,554,000
- MarCOP 34,166,000
- AbrCOP 53,078,000
- MayCOP 42,952,000
- JunCOP 58,840,000
- JulCOP 15,353,000
- AgoCOP 25,131,000
- Felipe NietoCOP 75,490,000
- Ana RestrepoCOP 58,514,000
- Diana SalasCOP 53,069,000
- Camilo OrtizCOP 42,447,000
Act 3
And this is how you ask it
Pick a question. You will see the query that runs and the figure it returns, and the chart above will light up where it belongs.
Pick a question to start
The query that ran
These six come prepared, which is why they answer instantly and at no cost. A company’s own agent answers its own questions — about its operation, querying its live database, not a copy.
Look at the second question
A client who stopped ordering appears in no chart. Charts show what happened; that question shows what is missing, and it is the kind of thing you hire an agent for rather than another dashboard.
What if this were your operation?
What you just saw is three pieces: a database that does not accept impossible data, a dashboard that updates itself, and an agent on top. They are built on what your company already has, without switching systems.