AI agents for New Relic
A Monmo AI agent works in New Relic the way a performance engineer would. It answers questions by writing and running NRQL, investigates open issues through APM traces and the errors inbox, compares response times before and after each deploy and keeps alert conditions useful. It works with the user you connect, reads freely and asks before changing a condition, a dashboard or an error's status.
Updated 2026-11-13 · Free to start. Paid plans from $15 a month.
What agents do in New Relic
Answer questions with NRQL
Ask which endpoints got slower this week or how many errors one customer saw. It writes the NRQL, runs it, checks the numbers are plausible and answers with the query shown.
AskWhich transactions got slower this week compared with last week? Show me the NRQL you used.
Investigate an open issue
When an alert condition opens an issue, it reads the incident, slow transaction traces, the errors inbox and recent changes for that entity, and writes what most likely caused it, ready to post in Slack once you allow it.
AskThere is an open critical issue on orders-service. What caused it, and is it still getting worse?
Compare before and after a deploy
Using the deployment markers in change tracking, matched to merges in GitHub, it compares response time, throughput and error rate for the hour before and after each release, and flags any service that got worse.
AskAfter every production deploy, compare the hour before and after and tell me if any service degraded.
Work through the errors inbox
Each morning it groups new error groups by service and users affected, links each to the likely code path, and suggests an owner with a draft Jira ticket. Assigning or resolving asks first.
AskEvery weekday at 9, go through new error groups in the inbox and suggest an owner for each.
Tune alert conditions
It finds conditions that opened the most incidents last month without leading to any action, and drafts new thresholds or durations from the real data. Editing a condition asks first.
AskWhich alert conditions opened the most incidents last month that nobody acted on? Propose better thresholds.
Build a service dashboard
It drafts a dashboard of NRQL widgets for a service, such as latency by endpoint, error rate and the slowest database calls. Creating it in your account asks first.
AskBuild a dashboard for the checkout service with latency by endpoint, error rate and slowest queries.
Report on service levels
Every Monday it reads your service levels, notes which ones missed their target and why, and writes the summary in Google Slides for the engineering review.
AskEvery Monday, put last week's service level results into three slides for the engineering review.
How it connects
- 1You connect New Relic once with a user key or your own sign-in. The agent works within the accounts and roles that user has, so a user with read-only access cannot change alerts even after an approval.
- 2You choose which agents may use New Relic. An engineering agent needs it, and a data agent may read it for reports; agents that handle sales or support do not.
- 3Running queries and reading traces, issues and dashboards happens without asking. Editing an alert condition or policy, creating a dashboard, muting, or assigning and resolving errors asks first.
More in Developer and IT
Other apps of this kind that agents work in, each with its own page.
Questions people ask
Can an AI agent write NRQL queries?
Yes. Ask in plain words and the agent writes the NRQL, runs it and explains the result, with the query shown so you can reuse it in a dashboard or an alert. If the first result looks wrong, for example a time window that misses the incident, it says so and adjusts the query.
Can it find out why our app slowed down?
It can do the digging. The agent compares the slow period with a normal one, reads transaction traces and database timings, checks deployment markers and points to the change most likely at fault. It works through the New Relic connection, or in its own browser when the connection does not cover something.
Will it change New Relic alert conditions without asking?
No. Every edit to a condition, policy or notification shows a card with the old and new settings and the data behind the change. Reading and querying are the only things it does without asking, so your alerting stays exactly as your team set it until you allow a change.
Is a New Relic user key safe to give an agent?
A user key acts with the permissions of the user it belongs to, so create one for a user with only the access the work needs. Approvals add a second check on any change. You can delete the key in New Relic at any time, and the agent loses access straight away.