Tap in to data.
Countdat turns anything you count into answers you can trust. Tap it, import it or pull it from your own systems, then ask it anything in plain English.
- Every answer shows its working
- Read-only where it reaches out
- Nothing is ever overwritten
From a tap to an answer.
Most of what a small team needs to know is a count of something. Countdat gives each one a counter, then answers your questions about it.
Count it
Tap a link on any phone, say a sentence, import a file, or let a scheduled job pull rows in from your own database.
Ask it
Ask in plain English. The answer comes back as a page: the number, what it means, and the charts that show it.
Trust it
Every answer shows its working: the exact queries that produced it, one click away, ready to copy or download.
Most AI answers ask to be believed. Ours can be checked.
Every answer page has a Show the working link. It opens the exact SQL the analyst ran, with a sentence on why each query ran, how many rows it read and how long it took.
| Table | Headroom | Used |
|---|---|---|
| orders | 71% | |
| events | 48% | |
| sessions | 22% |
-- Answer Log -- Counter: key-headroom · closest-to-full -- Work: 5 queries · 9 rows read · 1,919 ms -- 1. Finding the current table closest to full. -- 1 row · 180 ms SELECT table_name, column_name, ROUND(used_pct, 1) AS used_pct FROM rows WHERE checked_at = (SELECT MAX(checked_at) FROM rows) ORDER BY used_pct DESC LIMIT 1; -- 2. Comparing every table at the latest check. -- 9 rows · 212 ms SELECT ...
A colleague, an auditor, or anyone who reads SQL can see exactly what was asked of the data.
When the answer can't find a figure in the data, it says so rather than making one up.
The whole log is one script you can copy, or download as a .sql file.
One product. A café, an ops team and a fleet of servers.
Countdat has no built-in idea of coffee or databases. Each counter carries its own definition, so it fits whatever you count.
Coffee Count
People tap a shared link on their phone each time they pour a cup.
Key Headroom
Every hour, a job asks each database how close its keys are to their ceiling.
Weekly sales
Sales from the spreadsheets and emails you already have.
Protected by design, not by policy.
Four properties are built into how Countdat works. They are why you can point it at real systems.
Read-only where it reaches out
A job reading your database cannot change it - enforced three ways, on every query. A probe in pilot sends no query at all.
A preview before anything is saved
Imports, jobs, alerts, buttons and field changes all show their result first. Nothing is saved until you say yes.
A record of everything
Nothing is overwritten or deleted. A correction is a new row pointing at the old one; a delete is a void. Every row says where it came from.
Secrets stay secret
Passwords are encrypted at rest, used only by the server, and never shown to an AI model or written to a log.
Other tools show you data. Countdat answers your question about it - and shows you how.
| Instead of | Countdat |
|---|---|
| a spreadsheet | a counter collects its own rows - from taps, files and jobs - and keeps a record of every change |
| a configured report | ask the question you have today; the answer is written fresh from the data, not set up once and left |
| a chatbot over your data | answers are pages you can share and cite, with the queries behind them one click away |
| giving a tool your database password | jobs are read-only, and probes in pilot need no login at all |
Questions, answered.
Do I need to know SQL?
Can Countdat change my database?
Where does my data live?
What if the answer is wrong?
What happens when the data changes?
Who can tap my counter?
Is anything ever deleted?
Count anything. Ask it everything.
Make a counter, share the link, and ask your first question in a few minutes.