Source: https://internal-agents.com/agents/uber-ureview

# Uber — uReview

An event-driven AI code reviewer for Uber's internal review platform that generates, grades, filters, deduplicates, and posts findings while leaving engineers in control of the reviewed change.

- Approach type: Background agent
- Deployment stage: Scaled
- Autonomy: Drafts reviewed
- Evidence strength: Detailed primary
- Status: Internal
- First reported year: 2025
- Work: Code review
- Interfaces: Internal ui, Ci
- Invocation: Background, Event driven
- Entry reviewed: 2026-09-09

## Where people stay involved

- **pull request → filtered AI review findings** — Work product review · Level 3

## Overview

### Summary

An event-driven AI code reviewer for Uber's internal review platform that generates, grades, filters, deduplicates, and posts findings while leaving engineers in control of the reviewed change.

Fact · Reported · High confidence · `uber-ureview--summary`

Confidence reason: A linked first-party source states the claim.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Introduction; How It Works

## How it works

### Supporting component

Standard, best-practices, and AppSec reviewers generate findings for different issue classes

Fact · Reported · High confidence · `uber-ureview--primitives-0`

Confidence reason: A linked first-party source states the claim.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Comment Generation by Specialized Assistants

### Supporting component

Confidence grading, semantic deduplication, and historically low-value category suppression reduce noise

Fact · Reported · High confidence · `uber-ureview--primitives-1`

Confidence reason: A linked first-party source states the claim.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Post-Processing and Quality Filtering

### Supporting component

Developer ratings, addressed-comment detection, and a curated benchmark tune prompts, thresholds, and models

Fact · Reported · High confidence · `uber-ureview--primitives-2`

Confidence reason: A linked first-party source states the claim.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Evaluation and Continuous Improvement

## Supervision evidence

### Operating model assessment

Level 3 for pull request → filtered AI review findings; human attention boundary: work-product-review.

Inference · Catalog judgment · High confidence · `uber-ureview--operating-models-0`

Confidence reason: The source describes AI-generated review findings while engineers remain in control of the change.

Qualifications:

- Observation date: 2025-08-12

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md)

## Implementation details

### Harness

A prompt-chained pipeline separates comment generation, confidence grading, validation, semantic deduplication, and category filtering; three specialized assistants were in operation when published

Fact · Reported · High confidence · `uber-ureview--architecture-harness`

Confidence reason: A linked first-party source states the claim.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · How It Works; Comment Generation; Post-Processing

### Model

Periodic benchmark evaluation; Claude 4 Sonnet as generator with o4-mini-high as grader was the highest-F1 reported pairing

Fact · Reported · High confidence · `uber-ureview--architecture-model`

Confidence reason: A linked first-party source states the claim.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Empirical Model Evaluation

### Interfaces

internal-ui, ci

Fact · Reported · High confidence · `uber-ureview--architecture-interfaces`

Confidence reason: A linked first-party source states the claim.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Comment Delivery and Feedback Collection; Impact and Evaluation

### Tool access

Reviews eligible code in Uber's six monorepos across Go, Java, Android, iOS, TypeScript, and Python; richer internal artifacts were not yet connected when published

Fact · Reported · High confidence · `uber-ureview--architecture-tool-access`

Confidence reason: A linked first-party source states the claim.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Impact and Evaluation; Better at Catching Bugs than Assessing System Design

### Knowledge

Surrounding source context plus a shared registry of Uber-specific coding and style rules

Fact · Reported · High confidence · `uber-ureview--architecture-knowledge`

Confidence reason: A linked first-party source states the claim.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Ingestion and Preprocessing; Comment Generation by Specialized Assistants

### Context management

Comments and metadata are streamed through Kafka to Hive for feedback analysis, experiments, and operational dashboards

Fact · Reported · High confidence · `uber-ureview--architecture-context-mgmt`

Confidence reason: A linked first-party source states the claim.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Comment Delivery and Feedback Collection

## Reported metrics

### Headline claim

Uber's introduction reports reviews of over 90% of approximately 65,000 weekly diffs, with over 75% usefulness and over 65% addressed comments; a later paragraph says 65,000 diffs per month

Metric · Reported · Low confidence · `uber-ureview--headline-metric`

Confidence reason: The opening paragraph reports approximately 65,000 weekly diffs, but the cost discussion says 65,000 per month. These conflicting periods remain unresolved; neither is independently verified.

Qualifications:

- Reported by: Uber
- Scope: Introduction's reported weekly diff coverage, plus comment usefulness/addressed rates; later paragraph's monthly period conflicts and remains unresolved
- Denominator: Introduction reports approximately 65,000 weekly diffs; cost paragraph says monthly. Usefulness covers rated comments; addressed rate covers posted comments
- Method: Production coverage, developer ratings, and automatic addressed-comment detection
- Observation date: 2025-08

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Preserved content.md, lines 34, 98
- Contradicts · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Preserved content.md, lines 120 (65,000 per month, versus weekly in line 34)

### Key observation

Introduction reports reviews of over 90% of approximately 65,000 weekly diffs; cost discussion instead says 65,000 diffs per month

Metric · Reported · Low confidence · `uber-ureview--key-metrics-0`

Confidence reason: The opening paragraph reports approximately 65,000 weekly diffs, but the cost discussion says 65,000 per month. These conflicting periods remain unresolved; neither is independently verified.

Qualifications:

- Reported by: Uber
- Scope: Introduction's reported weekly diffs analyzed; monthly period in the cost paragraph conflicts and remains unresolved
- Denominator: Approximately 65,000 weekly diffs according to the introduction; same volume described as monthly in the cost paragraph
- Observation date: 2025-08

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Preserved content.md, lines 34
- Contradicts · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Preserved content.md, lines 120 (65,000 per month, versus weekly in line 34)

### Key observation

Over 75% of comments rated useful by engineers who interact with the tool

Metric · Reported · Medium confidence · `uber-ureview--key-metrics-1`

Confidence reason: A linked first-party source states the claim.

Qualifications:

- Reported by: Uber
- Scope: Comments rated useful by engineers who provide feedback
- Denominator: Comments with engineer interaction
- Method: Useful / Not Useful rating links
- Observation date: 2025-08

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Preserved content.md, lines 76, 98

### Key observation

Over 65% of posted comments addressed in the same changeset

Metric · Reported · Medium confidence · `uber-ureview--key-metrics-2`

Confidence reason: A linked first-party source states the claim.

Qualifications:

- Reported by: Uber
- Scope: Posted comments considered addressed in the same changeset
- Denominator: Posted comments
- Method: Five reruns on the final commit and semantic-similarity matching
- Observation date: 2025-08

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Preserved content.md, lines 82, 98

### Key observation

Median review latency of 4 minutes across all six Uber monorepos

Metric · Reported · Medium confidence · `uber-ureview--key-metrics-3`

Confidence reason: A linked first-party source states the claim.

Qualifications:

- The source does not report the denominator of this figure.
- Reported by: Uber
- Scope: Reviews across all six Uber monorepos
- Method: Production latency telemetry
- Observation date: 2025-08

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Preserved content.md, lines 94

### Key observation

Approximately 1,500 developer hours reportedly saved per week, based on an assumed 10-minute second review per processed commit

Metric · Reported · Low confidence · `uber-ureview--key-metrics-4`

Confidence reason: The source estimates about 1,500 hours using over 10,000 commits and a 10-minute assumption; the rounded figures do not arithmetically reconcile exactly and are not observed time savings.

Qualifications:

- Reported by: Uber
- Scope: Processed commits excluding configuration files; modeled second-review time savings
- Denominator: Over 10,000 commits per week
- Method: Processed commits multiplied by an assumed 10 minutes for a second human review
- Observation date: 2025-08

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Preserved content.md, lines 104

## Lessons and interpretation

### Lesson

Prefer fewer high-confidence findings over high comment volume

Inference · Catalog judgment · High confidence · `uber-ureview--lessons-learned-0`

Confidence reason: The article states this lesson directly and documents the associated mechanisms.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Precision Is More Valuable than Volume

### Lesson

Combine prompts with deterministic filtering, deduplication, evaluation, and feedback instrumentation

Inference · Catalog judgment · High confidence · `uber-ureview--lessons-learned-1`

Confidence reason: The article states this lesson directly and documents the pipeline.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Guardrails Are Just as Important as Prompts

### Lesson

Roll out gradually by team and assistant while tracking precision, recall, usefulness, and false positives

Inference · Catalog judgment · High confidence · `uber-ureview--lessons-learned-2`

Confidence reason: The article states this lesson directly and describes the rollout telemetry.

Evidence:

- Supports · [1] [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/) · [Preserved copy](https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md) · Trust Grows with Gradual Rollout

## Related implementations

- Related implementation: [Uber — Internal coding agent (unnamed)](https://internal-agents.com/agents/uber-coding-agent)

## Sources

1. [uReview: Scalable, Trustworthy GenAI for Code Review at Uber](https://www.uber.com/us/en/blog/ureview/)
   - Engineering blog · First party · Evidence
   - Original URL: <https://www.uber.com/us/en/blog/ureview/>
   - Published: 2025-08-12 · Accessed: 2026-08-13 · Last verified: 2026-08-31
   - Preserved copy in the repository: <https://github.com/steel-experiments/internal-agents-map/blob/main/archive/sources/uber-ureview-source-1/content.md>