Arize AI

ML and LLM observability platform with OpenTelemetry-based tracing, real-time evaluations, and agent observability for multi-step AI agent traces in production. Strong OTEL integration; Arize AX platform supports agent observability; Phoenix open-source library for LLM tracing; Series B funded; used by Uber, Square, and Chime.

Arize's OTEL-based approach aligns with the broader observability industry standard, making it a natural bridge between traditional APM and AI-specific monitoring for agent-driven applications.

arize.com Observability and AI Monitoring

United Statesprivate

At a glance

Total raised

$131M

last round series c

Headcount

51-200

approximate · public range

Headquarters

Berkeley

United States

Founded

2020

6 years active

Status

Private

Observability and AI Monitoring

Patents on file

USPTO grants

For subscribers

Subscribe to see the full profile

Including Funding rounds, Bull / Bear thesis, Stock + earnings, Roster changes, Patents, News, and Open roles.

Founding

50% off

/month

94 of 100 spots left

Full

$10

/month

Available once all 100 Founding Member spots are claimed.

Get full access

Already subscribed? Sign in →

Funding history

Total raised

$131M

Rounds tracked

Latest stage

Series C

Date	Stage	Amount	Valuation	Lead investors
Feb 20, 2025	Series C	$70M	—	Adams Street Partners
Sep 8, 2022	Series B	$38M	—	TCV
Sep 28, 2021	Series A	$19M	—	Battery Ventures
Feb 18, 2020	Seed	$4M	—	Foundation Capital

Founder backgrounds

Aparna Dhinakaran

Co-founder & Chief Product Officer

Co-founded Arize AI after working as an ML engineer and leader at Uber, Apple, and TubeMogul (acquired by Adobe). At Uber, she helped build Michelangelo, the company's core ML infrastructure platform. She holds a BS in Electrical Engineering & Computer Science from UC Berkeley and was a PhD candidate in Computer Vision at Cornell, where she grew frustrated that she couldn't answer basic questions about her models' production performance — the insight that led directly to Arize. Named to Forbes 30 Under 30. A former contestant on *The Amazing Race*, she has noted that the show's high-pressure, real-time decision-making felt surprisingly relevant preparation for startup life.

Jason Lopatecki

Co-founder & CEO

Co-founded TubeMogul, a video advertising platform, and led it as Chief Innovation Officer through its IPO and eventual acquisition by Adobe. Earlier, founded Illuminex, a video analytics startup acquired by TubeMogul in 2008. Holds a B.S. in Electrical Engineering and Computer Science from UC Berkeley. This second-time founder's experience scaling a data-intensive advertising platform from zero to public exit directly informs Arize AI's mission in ML and LLM observability — both companies sit at the intersection of large-scale data pipelines and production reliability.

AI bull / bear

Bull case

Arize occupies the convergence point between traditional APM and AI-native monitoring through its deep OpenTelemetry integration. OTEL has become the undisputed industry standard for observability data collection, and Arize's decision to build on it means enterprises can onboard without ripping out existing monitoring stacks. The February 2025 Series C — a $70 million round led by Adams Street Partners bringing total funding to $131 million — provides ample runway to capture the agent observability wave before it fully breaks. The product portfolio is strategically layered: Phoenix, the open-source LLM tracing library, drives bottom-up developer adoption and creates a pipeline into the commercial Arize AX platform. Meanwhile, AX's multi-step agent trace capabilities address the hardest unsolved problem in production AI — understanding what a chain of LLM calls actually did and why it failed. Enterprise logos like Uber, Square, and Chime validate that Fortune 500 engineering teams trust Arize for production workloads. As AI agents proliferate from pilot to production, the observability requirement shifts from nice-to-have to mandatory, and Arize's OTEL-native architecture positions it as the path-of-least-resistance choice for platform teams who already speak OpenTelemetry.

Bear case

The AI observability market is still undefined enough that Arize faces threats from every direction. On one flank, incumbent APM giants Datadog, Dynatrace, and New Relic are aggressively layering LLM monitoring into their platforms. These competitors already own the observability budget inside most enterprises and can bundle AI monitoring at marginal cost, making a standalone vendor like Arize expendable during vendor consolidation cycles. On the other flank, Arize competes with AI-native peers — LangSmith, Weights & Biases, and Braintrust — each with loyal developer followings and their own open-source beachheads. The Phoenix library, while a smart adoption wedge, also creates a ceiling: teams satisfied with basic LLM tracing may never convert to the paid AX platform. The $131 million raised implies a valuation that demands a large exit, but the total addressable market for dedicated AI observability remains speculative. Agent architectures are evolving rapidly, and if the dominant patterns shift toward embedded observability within agent frameworks themselves — rather than external platforms — Arize's standalone value proposition weakens substantially. Customer concentration risk also looms with only a handful of named enterprise logos.

Generated by deepseek-v4-pro · May 2, 2026Refreshes quarterly · see disclosures

In the news

Arize AI · last 30 days

No articles ingested yet for Arize AI. Once the hourly news pipeline is live, every article the classifier tags as mentioning this company appears here with its one-line AI summary and sentiment.

Watch

Leadership

Jason Lopatecki

Co-founder & CEO & Chief Executive Officer

—

Aparna Dhinakaran

Co-founder & Chief Product Officer

Michael Schiff

Chief Technology Officer

Products

Arize AX
Arize AX is the flagship AI & Agent Engineering platform that provides end-to-end observability, evaluation, and debugging for generative AI applications and agents in production. Built on OpenTelemetry-based tracing, it gives AI engineers real-time visibility into multi-step agent workflows, LLM calls, and retrieval-augmented generation pipelines. The platform includes LLM-as-a-judge evaluations, automated guardrails, drift monitoring, and performance analytics. It differentiates through its unified approach to development and production monitoring, enabling teams to iterate on prompts, evaluate outputs at scale, and debug production issues with span-level trace visibility across the entire AI stack.
Arize Phoenix
Arize Phoenix is an open-source LLM observability and evaluation library that enables developers to trace, visualize, and evaluate LLM applications during development and production. It provides tools for tracing LLM calls, visualizing complex decision-making through span trees, clustering embeddings to detect problematic responses, and running evaluations using LLM-as-a-judge techniques. Phoenix integrates with major LLM frameworks including LangChain, LlamaIndex, and OpenAI, and supports both Python and TypeScript SDKs. It differentiates itself as a free, community-driven alternative that gives teams full visibility into model behavior without vendor lock-in.
Alyx
Alyx is an AI engineering agent built directly into the Arize AX platform that autonomously automates complex AI engineering workflows. It can reason across multi-step processes to perform error analysis, prompt optimization, evaluation execution, and root-cause debugging of AI applications. Alyx acts as an intelligent copilot for AI engineers, helping them build, test, and ship reliable AI agents faster by handling repetitive tasks such as analyzing trace data, suggesting prompt improvements, running evaluations, and identifying regression causes. It differentiates by being purpose-built for AI engineering workflows rather than general-purpose coding assistance.

Most recent patents

ARIZE AI, INC.·2024-11-12
Systems And Methods For Optimizing A Machine Learning Model
US12141670
Abstract
Systems And Methods For Optimizing A Machine Learning Model
US12141670
Abstract A system for optimizing a machine learning model. The machine learning model generates predictions based on at least one input feature vector, each input feature vector having one or more vector values; and an optimization module with a processor and an associated memory, the optimization module being configured to: create at least one slice of the predictions based on at least one vector value, determine at least one optimization metric of the slice that is based on at least a total number of predictions for the vector value, and optimize the machine learning model based on the optimization metric.
View on Google Patents
ARIZE AI, INC.·2024-08-06
Data Drift Impact In A Machine Learning Model
US12056586
Abstract
Data Drift Impact In A Machine Learning Model
US12056586
Abstract Techniques for determining a drift impact score in a machine learning model are disclosed. The techniques can include: obtaining a reference distribution of a machine learning model; obtaining a current distribution of the machine learning model; determining a statistical distance based on the reference distribution and the current distribution; determining a local feature importance parameter for each feature associated with a prediction made by the machine learning model; determining a cohort feature importance parameter for a cohort of multiple features based on the local feature importance parameter of each feature in the cohort; and determining a drift impact score for the cohort based on the statistical distance and the cohort feature importance parameter.
View on Google Patents

Open roles

View all careers ↗