What challenges did Baz face before AI implementation?

Code review was manual and ineffective because it checked compilation but not whether the delivered UI and behavior matched product requirements in Figma and Jira. QA teams spent hours manually validating preview environments, causing delays and regressions.

What was the impact of the AI solution?

Reduced reported bugs by up to 50%. Shifted feature verification earlier and automatically onto pull requests.

Source

Evidence: Low35/100

Baz automated AI code review/spec validation using Amazon Bedrock AgentCore

Use case typeDeveloper productivityUpdated Jun 2, 2026

Baz built a Spec Review agent to automate code review and product validation for software development workflows. The system checks whether implemented behavior matches requirements from Figma and Jira, not just whether code compiles.

Organization: Baz
Industry: Tech & Comms
Location: United States
Published: June 2026

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Reported bugs: 50% decrease

AWS Machine Learning BlogJun 2, 2026Blog postExplicit claimLow evidence strength

reducing reported bugs by up to 50%

Normalized claim

Time-to-merge: 30-70% decrease

AWS Machine Learning BlogJun 2, 2026Blog postExplicit claimLow evidence strength

time-to-merge by 30–70%

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Baz
Provider: AWS
Maturity: Unknown
Linked source: AWS Machine Learning Blog

No explicit deployment-stage evidence found.

Customer identity supportedSource describes one deploymentMaturity evidence evaluated

Primary read

Use case focus

Showing 2 of 2

1Developer productivity
2Workflow automation

Baz built a Spec Review agent that, on GitHub pull request triggers, queries Figma and Jira, decomposes specs into requirements, and uses Amazon Bedrock-powered subagents to validate the live preview environment.
The agent uses Amazon Bedrock AgentCore Browser Tool for isolated browser automation, compares UI behavior and states to design and acceptance criteria, and posts results to GitHub PRs, Slack, and Jira.

Technologies

Amazon Bedrock Amazon Bedrock AgentCore Amazon EKS GitHub Slack Jira Figma

Reduced reported bugs by up to 50%.
Shifted feature verification earlier and automatically onto pull requests.

Architecture

A GitHub webhook or manual trigger starts an Amazon EKS-hosted Baz Spec Review agent. The agent queries Figma via MCP and Jira via REST APIs, decomposes requirements into visual and functional checks, then launches isolated subagents that use Amazon Bedrock for reasoning and Amazon Bedrock AgentCore Browser Tool for secure browser-based validation of the live preview environment. Findings are consolidated and sent back to GitHub pull request comments, Slack, and Jira.

Sources & evidence1

Evidence: Low35/100Evidence strength