What AI spend insights does Cognocient generate automatically?

Proactive cost intelligence from your AI Cost Advisor, generated nightly. Warnings, opportunities, and patterns surfaced automatically — no questions required.

Cognocient's AI Cost Advisor generates proactive insights every night from your call data — cost warnings, optimization opportunities, and unusual patterns surfaced before you think to ask. Insights appear in the dashboard and optionally in Slack or email.

Insights require a connected AI provider key. The same key that powers the AI Advisor generates nightly insights. See AI Advisor for setup.

What are Insights?

While the AI Advisor answers questions you ask, Insights are proactive. Every night, your AI Cost Advisor analyses your spend data and generates a prioritised list of findings — without you having to ask. These appear as cards in Intelligence → Insights and as a banner alert on the Executive View.

Insights fall into three categories:

Type	Colour	What it means
Warning	Red	A problem that is costing you money right now — a runaway spend spike, a feature trending over budget, or a repeated pattern of waste.
Opportunity	Green	A specific action you can take to reduce cost or improve efficiency — caching candidate, model downgrade, or unused budget room.
Pattern	Purple	A trend that is not urgent today but worth monitoring — seasonal usage, model cost changes, or evolving team behaviour.

Each Insight also has an urgency tag (High or Medium). High-urgency insights from the previous night appear in the purple banner on the Executive View — the "N new insights from your AI Cost Advisor" notification.

What an Insight looks like

Each card shows:

Headline — one-sentence summary of the finding
Detail — 2–4 sentences with specifics: which feature, what changed, by how much, over what period
Action — what Cognocient recommends you do next

Example warning:

Headline: document-extractor spend up 182% vs. last month
Detail: November spend on document-extractor was $2,512 vs. $890 in October — a 182% increase. Call volume tripled after your v2.1 deployment on Nov 3rd and average token count per call increased 47%.
Action: Enable prompt caching for document-extractor (68% prefix similarity detected). Estimated saving: $840/mo.

Acting on Insights

Two actions are available per insight:

Button	What it does
Resolved	Marks the insight as acted on and removes it from the list. Use this when you have taken the recommended action or deliberately accepted the situation.
Dismiss (×)	Removes the insight without marking it as resolved. Use this when the insight is not relevant to you.

Both actions remove the insight from your active list. Your AI Cost Advisor will surface new insights nightly if the underlying condition persists.

Filtering Insights

Use the filter chips above the insight list to focus:

Filter	Shows
All	Every active insight
Warning	Spend problems requiring attention
Opportunity	Cost-saving actions available
Pattern	Trends to monitor
High Urgency	Any insight flagged urgent, regardless of type

How Insights are generated

The nightly insight run:

Fetches your last 30 days of spend data (token counts, models, features, costs)
Compares current week vs. prior 3-week baseline
Runs waste detection and anomaly scoring
Calls your connected AI provider (Anthropic → OpenAI → Azure) to synthesise findings into plain-English cards
Stores results; the Executive banner updates on next login

The typical run produces 3–8 insights per night. Your provider key is used for the synthesis step — cost is approximately $0.01–$0.03 per nightly run.

If you have no insights after the first night, check that your provider key is active in Settings → API Keys. If no key is connected, the nightly run is skipped.

Accessing Insights

Two entry points:

Intelligence → Insights in the left sidebar
Executive View → click View on the insight banner (only appears when there are unread high/medium urgency insights)

Related: AI Advisor · Anomaly Detection · Executive View