Skip to content
ai-agent.review — AI Agent Reviews, home
← All news

Product updates

New Relic announces AI Evaluation preview for November

Vendor-reportedChecked: 2026-10-07[release]

Editorial attribution: Gosia Madej

Prepared with AI assistance. Editorial attribution does not imply personal product testing or manual review of every story.

Published: · Event date:

The planned feature links AI response scores to application traces. Support teams should distinguish sampled evaluation from prevention.

Editorial assessmentChecked: [release]

A conversation trace branches toward a magnifying glass, while the original path continues to a customer reply.
Original illustration of a separate evaluation path; not a New Relic interface or evidence of product performance. · AI Agent ReviewsImage use and permissions

New Relic announced AI Evaluation on October 6, with public preview planned for November. It describes asynchronous LLM judging attached to traces and checks on sampled inputs and outputs.

Vendor-reportedChecked: [release]

Its accompanying product update separately describes agent monitoring for tool calls and handoffs, plus OpenTelemetry support. These are related observability capabilities, not evidence that the announced evaluation preview is already available.

Editorial assessmentChecked: [roundup]

Existing monitoring documentation exposes a practical tradeoff: disabling AI content recording also stops automatic token-count forwarding; applications can supply counts through a local callback. This is existing instrumentation guidance, not confirmed configuration for the forthcoming preview.

Editorial assessmentChecked: [docs]

Checked against sources

AI-assisted source check

AI-assisted checking compared the dated announcement, product update and instrumentation documentation. All are vendor sources; no product test, independent performance verification or human review was performed.

Editorial assessmentChecked: [release][roundup][docs]

The announcement does not establish per-response blocking or specify preview pricing.

Editorial assessmentChecked: [release]

Source evidence

Original source excerpt. This shows what the source says; it does not establish product performance.

AI Evaluation will be available in public preview in November.

Read original source [release]

What changes for users

For teams operating support bots, plan a pilot that links customer-reported failures to tool calls and handoffs. Ask which data the evaluator needs when content recording is disabled, and test whether token-cost reporting survives your privacy settings. Keep customer feedback as a separate check on automated judgments.

Editorial assessmentChecked: [roundup][docs]

Sources