Dashboard AI Assistants — Insight Generator

Natural-language executive summaries, root-cause and anomaly detection, and actionable recommendations for chart-heavy dashboard screens.

The Insight Generator

Dashboard screens don't have rows to select — they have charts and metrics to interpret. The Insight Generator is the AI Assistant built for exactly that: instead of acting on data, it reads the dashboard's charts for you and turns them into plain-English findings. It runs on three pillars — a natural-language executive summary, root-cause and anomaly detection, and actionable recommendations — and it is arriving across TotalApp's chart-heavy screens now, starting with the MTBF/MTTR Reliability Dashboard.

1. Natural-Language Executive Summary

Instead of manually reading every chart on the dashboard, the Insight Generator produces a short, plain-English summary of what the numbers say right now.

Example — MTBF/MTTR Reliability Dashboard

"Over the last 12 months your average MTBF is 575h, above target — but CNC Machine A1 is the weak point at only 397h." One sentence replaces a full read-through of the reliability charts.

2. Root-Cause & Anomaly Detection

When a metric spikes or drops, the Insight Generator doesn't just flag the change — it explains which equipment, asset group, or segment is actually driving it, in context.

Example

"MTTR increased sharply in Q3 — driven mostly by the Boiler System H8 asset group." Instead of a lone chart showing a spike, you get the asset group responsible for it.

3. Actionable Recommendations

The Insight Generator closes the loop with a concrete next step tied directly to the data — not a generic tip, but a specific change with an estimated impact.

Example

"Boiler System H8 had 22 interventions in the last 12 months. Shortening the preventive maintenance interval to 2 weeks could reduce average MTTR by roughly 15%." A quantified recommendation, not just an observation.

Rolling Out Across Dashboards

The Insight Generator is arriving across TotalApp's dashboard screens, starting with:

Maintenance — Reliability Dashboard

MTBF/MTTR trend interpretation, asset-level root cause, and preventive-maintenance recommendations.

Manufacturing — OEE Dashboard

Availability, performance, and quality trend summaries with line- and shift-level root cause.

Other chart-heavy screens

Additional analytics dashboards across TotalApp are being wired up to the same Insight Generator pattern.

A note on paths and naming

Each dashboard's Insight Generator is scoped to that dashboard's own screen and its own metrics. Internally, a dashboard's path stays short and domain-meaningful — like the Reliability Dashboard's maint-reliability — rather than a generic /dash-style shorthand, so the assistant's scope is always obvious from the URL itself.

Frequently Asked Questions

Which dashboards have the Insight Generator today?

The MTBF/MTTR Reliability Dashboard in Maintenance Operations is the first live screen. The Manufacturing OEE Dashboard and other chart-heavy analytics screens are being wired up next as part of the ongoing rollout.

Does the Insight Generator change my underlying data?

No. Unlike the Table & List Assistant, the Insight Generator is read-only — it interprets charts and metrics and produces summaries, root-cause explanations, and recommendations, but never edits records itself.

How does root-cause detection know which asset or segment is responsible?

It correlates the metric that changed with the underlying breakdown data behind that dashboard's charts (e.g. per-asset MTTR figures) to identify which specific group is driving the overall trend, rather than just flagging that a change occurred.

Are the recommendations generic tips or specific to my data?

They're tied directly to the dashboard's own numbers — for example, a specific asset's intervention count and an estimated MTTR reduction if a maintenance interval is shortened, not a generic best-practice tip.

Which Writer Engine powers the Insight Generator?

The same Agentic configuration used across TotalApp — Ollama, In-Browser (ONNX/WebGPU), Cloud LLM, or Local CLI, set in Settings → Agentic. See AI Assistants → Overview for how this is shared with Table & List Assistants.