PersatePersate documentation

Alerts and risks

Standing watches over the legislative stream: define a topic, the platform monitors every Sejm speech, public-pulse post, recording, and uploaded file for matches, and surfaces them as trigger events.

Alerts and risks is the platform's standing-watch facility. Each alert defines a topic — typically a description in natural language and a set of keywords — and the platform monitors every incoming legislative signal for matches. When a match is found, a trigger event is recorded against the alert and surfaced in the user's view.

Alerts are accessed via the sidebar entry CORE → Alerts & Risks.

What an alert watches

An alert is matched against four content channels, each surfacing a distinct kind of event:

  • Sejm speech — transcribed speeches from analysed Sejm recordings.
  • Public pulse — posts captured from the X (Twitter) accounts of tracked stakeholders.
  • File — documents uploaded to the user's Repository.
  • Live — segments of currently broadcasting sessions, captured by the Record Analysis pipeline.

The same alert may match across all four channels simultaneously; each match is logged independently.

The alerts list

The view opens to a grid of alert cards. Cards are presented as a two-column grid on desktop viewports and a single column on narrow ones.

The alert card

Each card displays:

  • A banner image — an AI-generated thematic illustration produced when the alert is created. While the illustration is being generated the banner falls back to a neutral gradient.
  • A status chip in the lower-left of the banner — ACTIVE or PAUSED.
  • The alert's name as the heading.
  • A short description beneath the name.
  • A footer row showing Last trigger followed by the most recent date on which a matching event was recorded. Where no events have ever been recorded the value reads N/A.
  • An optional attribution popover in the footer reproducing the source attribution for the banner image.

Selecting a card opens its detail view, described in Inspecting an alert.

The right sidebar provides two controls.

Create alert

A prominent Create alert button at the top of the sidebar opens the alert-creation form. The full form is described in Creating and editing alerts.

A free-text field that filters the loaded alerts by name, description, or keyword. Matching is case-insensitive and updates as the query is typed.

Alert lifecycle

An alert moves through three states:

  • Active — the alert is matched against incoming events. New matches accrue as trigger events.
  • Paused — the alert is retained but no new matching is performed. Trigger events accumulated before the pause remain visible.
  • Deleted — the alert and its trigger history are removed permanently. Deletion requires confirmation and is irreversible.

State transitions are performed from the alert's detail view via the Pause / Resume controls and the Delete action.

Empty and error states

  • No alerts found — no alerts have been created on the account yet, or the active search excludes every loaded alert. The Create alert button in the sidebar is the entry point for the first.
  • A red error banner appears at the top of the list when communication with the feature service fails. The page may be reloaded; persistent failures are reported to the operations team automatically.

Pages in this section

  • Creating and editing alerts — the alert form: name, description, keywords, visibility, AI enhancement, and the duplicate-alert warning.
  • Inspecting an alert — the detail view: hero banner, action controls, trigger history, and the sharing panel.

On this page