> ## Documentation Index
> Fetch the complete documentation index at: https://docs.attensira.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Your first reading

> What happens the moment your subscription goes active: a dedicated pass reads every tracked prompt right away, and files what it finds in your Inbox.

Paying for Attensira starts a read, not a wait. The instant your subscription becomes active, every prompt you are tracking is read on every model your plan includes — on a dedicated fast lane that runs once, separately from the daily schedule that takes over afterward.

## What you see while it runs

The onboarding screen shows this pass live: one row per prompt on each model, resolving from waiting to read as the results come back. Nothing here is a placeholder — every row is a real, individually queued read, and the screen is watching the same record the API does.

Not every read lands cleanly. A model can time out, refuse a request, or return something we cannot parse. Reads like this are retried, and a row that could not be read after retrying **says so** rather than sitting there forever or quietly counting as a zero. "3 could not be read yet" is a fact about that row, not an estimate.

## When the setup pass reports in

You do not wait for every last row. The setup pass reaches a floor and reports what it has as soon as **at least half the reads are in**, or after **twenty minutes**, whichever comes first — a customer who paid a minute ago is owed an answer, even a partial one. What it finds is filed in your [Inbox](/agent/inbox), so the first thing this subscription does is hand you something to look at, not a spinner.

A full setup pass like this — reading every tracked prompt across the whole plan at once — runs **at most once a day per workspace**. Re-activating a subscription you had just turned on again the same day does not queue a second one.

## After that

Once the setup pass has run its course, your prompts move onto the same daily schedule as every other workspace: each prompt × model × country combination is read once per workspace-local day, at the sampling depth your plan sets. See [How we collect data](/measure/how-we-collect-data) for the full mechanics of that ongoing cadence.