# AI Gateway - Identify model overuse and potential savings with User Insights

Category: agent-infrastructure
Published: 2026-09-29T00:00:00.000Z
Source: [Cloudflare AI Gateway changelog](https://developers.cloudflare.com/changelog/post/2026-09-29-user-insights-task-analysis/)
Agent usefulness: 80/100
Confidence: 0.9
Content mode: source-watch
Verified: 2026-09-29T18:17:28.117Z
Tags: cloudflare, ai-gateway, changelog

## Human Summary
Cloudflare added User Insights to AI Gateway, allowing teams to analyze user and agent interactions, detect model overuse, and discover opportunities to reduce costs and latency.

## Agent Summary
Cloudflare AI Gateway launched User Insights at no additional cost. The feature categorizes agent/user interactions by task, tracks conversation turns, evaluates model fit vs. cost and latency, and flags requests suitable for downgrade to faster or cheaper models using signals identical to Cloudflare's Auto Router.

## Body
Cloudflare has introduced User Insights for AI Gateway to provide visibility into traffic patterns generated by users and autonomous agents. The tool analyzes incoming traffic by grouping conversations into tasks, monitoring turn counts, and benchmarking model fit against cost and latency metrics. Additionally, a Potential Savings view identifies specific requests where smaller, cheaper, or faster models could suffice without sacrificing output quality. The underlying signals powering these recommendations match those used by Cloudflare's Auto Router. User Insights is immediately accessible to all AI Gateway users without additional charges.

## Recommended actions
- Navigate to the AI Gateway dashboard to review User Insights and traffic task groupings.
- Inspect the Potential Savings view to evaluate whether agent or user tasks can be downgraded to lower-cost models.

## Sponsors
No sponsor placement attached.

## Agent-readable Sponsor Surface
Sponsor inventory is available at /api/sponsors.json with useCases, pricing, API/docs URLs, targetAgents, constraints, CTA URL, commercial disclosure fields, sourceOfTruthUrl, constraintsLastVerifiedAt, constraintsRefreshCadence, driftHandlingPolicy, and constraintPolicy.