APEX Project Eye 26.2 introduces Watchtower Summary, a new Watchtower feature that puts a compact summary of every scan directly on the dashboard, so you can see the state of a scan — and what to do about it — the moment it finishes.

Configuring the Summary in Watchtower Administration

To configure Watchtower Summary, go to Watchtower → Administration and open Summary. 

The summarization process runs either after a linked scanner run (e.g., APE Daily Scanner) or at a defined time interval. Each time it runs, it checks sensor readings recorded since the previous run and produces a summary of those readings.

Two summarization methods are available: AI (which requires an AI provider to be configured first) or non-AI (PL/SQL). By default, the feature is enabled using the non-AI method.

In the Watchtower Summary Configuration screen, set the Readings Limit and Sensors Limit, choose whether it executes after a Watchtower scan or at a defined time, and pick the notification channels that should receive the full summary. You can also preview the generated message with Summarization Message Preview.

Readings Limit and Sensors Limit

Readings Limit and Sensors Limit (both default to 5) are thresholds that control how much detail the non-AI summary shows — they have no effect on the AI-generated summary, long or short:

• Readings Limit: if the number of readings recorded since the last run is at or below this value, every reading is listed individually; above it, the summary falls back to a per-sensor rollup or a totals-only view.
• Sensors Limit: if the number of distinct sensors with readings is at or below this value, each sensor is listed with its severity counts, in alphabetical order; above it, only totals are shown.

Neither setting caps how much data the AI summary reads or sends to your AI provider. If you're using the AI method, both fields can be left at their defaults.

Two summarization methods: AI or non-AI (PL/SQL)

The Use AI to analyze sensor readings and produce message toggle controls how each summary is generated:

• No: the summary is produced by built-in PL/SQL logic, working directly off the sensor readings and thresholds.
• Yes: the summary is generated by an AI provider instead, producing a narrative read of the same data.

To use the AI version, connect an AI provider under Administration → AI Settings, then enable it for Watchtower. The AI-Powered Features screen shows Watchtower AI Summary and Watchtower AI Summary (Short) as ready to use once a provider and model are configured.

No AI provider configured yet? You can still bring AI into the loop manually: use Download LLM summary on the Scans page to export a scan as a Markdown (or JSON) file, then paste it into your own LLM chat for an AI-assisted read. 

What the summary includes

Each dashboard summary — AI or non-AI — is built to be scanned briefly, not read like a report. It includes:

• A status badge giving the overall health of the scan in one glance
• A one-sentence overview of what happened
• A color-coded sensor list, ranked by worst severity first, so the sensors that need attention surface immediately
• Up to three recommended actions, telling you what to look at next instead of leaving you to work it out from raw readings

Because it lives on the dashboard itself, there's no need to open an email or wait for a digest — the summary is there the moment the scan finishes, whether it's the AI-generated narrative or the built-in PL/SQL version.

Summary email

If you've configured notification channels, you'll also receive the full summary by email —delivered to your APE Inbox, your mailbox, or both, depending on what you've set up.