Optimal AI Uses the Gemini API to Cut Code Review Times by 50%

Code reviews, while crucial for quality, often become a bottleneck in fast-paced development. Optimal AI is changing that. Their mission: "give engineers back their time" using AI to automate engineering and compliance. Their solutions include Optibot, an AI code reviewer focused on security and compliance, and an insights platform powered by the Gemini API that optimizes development velocity.

Previously Optimal AI faced challenges with the speed and contextual understanding needed for truly effective AI code review. "The biggest challenge was contextual understanding—we needed a model that could look at code changesets and actually contextualize them," explains Syed Ahmed, Co-founder & CTO.

By integrating the Gemini API, Optimal AI has significantly enhanced its offerings:

• Boosted code review speed and accuracy: Optibot, powered by the Gemini API, automatically reviews pull requests for security vulnerabilities, compliance risks, and coding patterns, providing actionable feedback and dramatically reducing review times.
• Extracted actionable engineering insights: The Gemini models analyze data from GitHub and Jira to identify bottlenecks and understand engineering performance, effectively differentiating between productive activity and code churn.
• Balanced speed and sophistication: Optimal AI leverages Gemini 2.5 Pro for complex analysis and in-depth code understanding, while Gemini 2.0 Flash provides the speed needed for low-latency tasks like quick summarizations.

Optimal AI’s implementation showcases the Gemini API's flexibility:

• Models used::
  • Gemini 2.5 Pro: For in-depth code analysis, security checks, contextual feedback on pull requests, and identifying complex engineering patterns for performance insights.
  • Gemini 2.0 Flash: For low-latency tasks such as file tree scanning and generating quick summaries.
• Key features & implementation::
  • Contextual understanding: The Gemini models’ large context window is crucial for interpreting complex code changesets and understanding broader engineering patterns.
  • Multi-language support: The Gemini models’s improved ability to handle multiple programming languages and frameworks was a significant win for Optimal AI.
  • Google AI Studio: The team heavily uses Google AI Studio for rapid prompt testing, model evaluation, and iteration. "The ability to see outputs alongside the implementation code has made it much easier for our engineers to experiment," notes Ahmed.

The impact of integrating Gemini has been significant for Optimal AI and its customers. Key results include:

• 50% reduction in pull request cycle times: Engineers spend less time waiting for reviews and more time coding.
• Rapid customer adoption and expansion: Companies like MongoDB have significantly increased their usage of Optimal AI after experiencing its benefits, growing from 5 to over 40 engineers.
• Successful $2.25 million pre-seed funding round: This was achieved in private beta, largely driven by the traction and results demonstrated with the Gemini API-powered features.

"Teams love that Optibot helps them cut PR review times in half, freeing up engineers to spend more time coding instead of waiting on approvals," shares Ahmed.

Optimal AI is focused on expanding its suite of AI agents to automate even more repetitive tasks. They are currently developing "Code Radar," an agent designed to autonomously monitor, patch, and secure codebases. Reflecting on their journey with the Gemini API, Syed Ahmed offers this advice to fellow developers:

"Go straight to Google AI Studio—it has better tooling, better documentation, and makes experimentation a lot more efficient." He also emphasizes, "Take full advantage of the Gemini model’s context window. Feed the models as much relevant context as possible...the more context we provided, the better the AI’s reasoning became."

Optimal AI's success demonstrates how the Gemini API can transform software development, enabling teams to build better software, faster.

Ready to build? Explore the Gemini API documentation and get started with Google AI Studio today.