Back to Work

AIM Monitoring

Cost Management, Performance Optimisation, End-to-End Visibility, Model Comparison and Optimisation.

AI Monitoring overview

Project Story

In the world of AI chains, visibility isn't just about watching - it's about understanding, improving, and protecting. Every input, every output, every token, every millisecond, and every potential danger is recorded, analysed, and learned from. That's how we build AI systems worthy of trust.

Overview

AIM enables end-to-end tracing of LLM chains with visibility into input-output, errors, token usage, and latency at each step, along with robust output quality and security evaluations while ensuring accuracy and safety.

Timeline

Nov 2024 - June 2025

Launch

Rolled out to customer end on February event launch for LP.

Latency

78%

Reduction in Latency/Response Time

Throughput

100%

Increase in Throughput

Errors

54%

Reduction in Error Rates

Solution

New Relic turns AI agents from "hoping the AI works" to knowing exactly how it performs, what it costs, and where to improve. New Relic does not just monitor AI agents; it helps teams master them.

Problems

Business Problem

AI Monitoring fundamentally solves the lack of control, visibility, and confidence in deploying and operating AI applications, particularly those involving Large Language Models in a production environment.

User Problem

User problems boil down to a lack of observability, control, and confidence when operating LLM-powered applications, leading to wasted time, increased costs, and hallucination prompts. AIM provides the necessary tools to overcome these hurdles.

Users

Users responsible for building, operating, protecting, or benefiting from AI applications need monitoring.

AI/ML engineers persona

AI/ML Engineers

These users focus on the infrastructure and operational aspects of AI systems. They deploy and maintain ML models in production, and manage LLM models and scaling.

CXO persona

CXO

They need high-level dashboards showing AI initiative success, strategic KPIs, and overall AI program health.

Product and cost managers persona

Product / Cost Managers

They track conversion rates, user engagement, and feature adoption related to AI-powered functionality, identifying which AI features are driving business value.

Information Architecture

In AI monitoring, you're not just dealing with more data; you're dealing with fundamentally different data that requires intelligence to interpret. Without IA, you are drowning. With IA, you're surfing.

AIM information architecture

User Growth Statistics

Here are the number of users increased data about user growth, different AI monitoring usage patterns, clicks, emails, and click-through rates.

Final Design

View Figma Design

Our Customers Using AI Monitoring

New Relic's AI monitoring customers are 298 leads through organic email campaigns and landing page sign ups, leads generated from account executives, and telemetry. Since AIM preview supports Python and Linux, 88 customer profile fits can leverage AIM successfully.

AI monitoring customer 1
AI monitoring customer 2
AI monitoring customer 3

Appreciations in New Relic

New Relic's AI monitoring customers are 298 leads through organic email campaigns and landing page sign ups (214), leads generated from account executives (11), and telemetry (73). Since AIM preview supports Python language and Linux environment, there are currently 88 customer profile fits who can leverage AIM successfully.

Impact

214

Total views for every 60 min

43

Applications onboarded in a month