New Relic Adds Monitoring for AI Coding Assistants

New Relic, a software monitoring platform provider, has launched a new AI Coding Observability tool aimed at bringing oversight to an area that has so far been largely invisible to IT departments: the work of AI-based coding assistants. The solution went live on June 23, 2026, as an open source feature, available at no extra cost beyond New Relic's standard data processing rates.

Until now, tools like Cursor or GitHub Copilot have operated largely outside the traditional monitoring stack companies use to oversee production applications. Engineering teams have widely rolled out various coding assistants, but had no systematic way to check how much these tools actually cost, how they affect code quality, or what data leaves the organization while they work.

What the tool actually measures

AI Coding Observability extends New Relic's production monitoring to the coding stage itself. Instead of relying on anecdotal claims that a given assistant "speeds up the team," the tool is meant to provide hard data: productivity metrics tied to developers' actual output rather than to self-reported claims.

The second feature concerns costs. Companies using several coding assistants at once often don't know how much they're actually paying for the tokens consumed by individual teams. New Relic lets them track AI spending, forecast it against budgets, and send alerts before an organization exceeds a set threshold.

Data stays inside the company network

The third pillar is regulatory compliance. A local mode, described as zero-outbound, runs telemetry queries entirely within the organization's private network, which is meant to prevent snippets of code or prompts from leaking to public cloud models. This addresses concerns from security teams, who view unauthorized transfer of source code to external AI models as a serious risk to a company's intellectual property.

New Relic stresses that the tool is open source and source-available, which is meant to let security teams independently verify how it works. Basing it on the open OpenTelemetry and Model Context Protocol standards is meant to ensure vendor neutrality: a company isn't locked into a single coding assistant and can switch between tools at any time without losing the telemetry data it has collected.

You can't manage what you can't see. AI coding assistants have a measurable impact on the business, but without real-time oversight of how they behave, organizations are scaling risk just as fast as they're scaling production. - Brian Emerson, Chief Product Officer, New Relic

Scale of the problem

The company cites a Gartner forecast according to which, by 2028, as many as 90 percent of software engineers at enterprises will be using AI-based coding assistants. At that scale of adoption, a lack of visibility in this area stops being a minor oversight and becomes a real gap in managing IT risk and costs.

For finance departments, this means an end to AI subscription bills showing up as a surprise line item at the end of the quarter. For security teams, it's a way to answer the question of which AI tools have access to production code and whether data leaves company infrastructure at all.

What it means for Polish companies

Polish development teams, like the rest of the market, are increasingly using several coding assistants at once, often without central records of who uses them and to what extent. Solutions like AI Coding Observability address the same problem flagged by earlier reports about companies losing control over AI spending or employees using AI tools without their managers' knowledge. For companies subject to EU data protection requirements, the local mode, which keeps code telemetry inside their own infrastructure rather than sending it to external services, may be especially relevant.

Adopting the tool doesn't require companies to change the coding assistants they use - New Relic normalizes telemetry data from five different products into a single dashboard, which is meant to make it easier to compare their effectiveness and costs without having to migrate between tools.