Tranzzo has been included in the Google Cloud EMEA SMB AI Playbook with its Code Review automation case study.

Tranzzo has become part of "The SMB AI Playbook: Driving Real Value in EMEA" — an initiative from Google Cloud that brings together leading small and medium-sized businesses in the region that are implementing artificial intelligence for real business impact.

Our Code Review Automation case study was featured in the publication as an example of how AI can accelerate development, improve code quality, and relieve engineers of routine checks.

For Tranzzo, this is not just an award. It is confirmation of our strategy: innovation must bring measurable value. And code review automation has become just such an example.

🔗 Whitepaper доступний у розділі “How businesses are using AI across EMEA today” EMEA SMB AI від Google Cloud.

Tranzzo is a global online payment platform that serves over 3,000 active customers and processes over 10 million transactions every month. Supporting operations in the financial sector, which is regulated by the PCI DSS standard, we cannot afford to compromise on code quality. Every mistake or missed vulnerability is not just a bug, but a potential financial risk.

With the growing number of teams and the increasing complexity of our software architecture, the classic code review process began to slow down development. Senior engineers spent too much time manually reviewing pull requests, and repetitive tasks were exhausting the team.

A problem typical for scalable products has arisen:

How to maintain code quality and security while accelerating development speed?

The answer came via AI. Instead of relying solely on human reviews, the team decided to integrate LLM (Large Language Model) into the development process so that the code could be analyzed automatically — even before human intervention.

The system uses Gemini 2.5 Pro via API. This allows AI to be integrated into the existing CI/CD process without changing the developers' tools.

AI Code Reviewer is automatically activated when you create a pull request in Bitbucket. It analyzes:

• changes in the code (diff);
• context of the task from the Jira description;
• availability of tests;
• compliance with internal rules (via CODE_REVIEW.md);
• potential security or pattern violations.

Output — a structured report that is sent to the developer for manual review, with comments directly in the pull request, where each item has a category: bug, security, performance, architecture, testing, code quality, business requirement.

Stages of implementation

Phase 1. Pilot (1–2 weeks)

We started by building a custom workflow integrated with our development tools.

Key features:

• automatic launch of AI Code Review after creating a pull request (PR);
• reading the context of the task from the ticket description;
• Integration with LLM via API — Gemini 2.5 Pro was chosen for its accuracy and optimal cost.

Prompt engineering was aimed at identifying missed implementations, gaps in testing, code smells, and potential vulnerabilities.

AI comments were automatically structured in PR, and the pilot team tested the process in a real environment.

Phase 2. Context expansion (weeks 2–3)

Based on feedback, we have expanded the system's capabilities:

• added git checkout for full project context;
• optimized diff file processing;
• introduced CODE_REVIEW.md with rules for each project;
• taught AI to take into account architectural patterns and test coverage.

Now, AI analyzed not only changes in the code, but also the entire structure of the repository, which significantly improved the accuracy of recommendations.

The suggestions were classified by type of impact — from preventing errors to improving test quality.

Phase 3. Scaling (3+ weeks)

The system was deployed across all teams without any changes to processes. Now, every pull request automatically undergoes AI review before human review.

Results

Quantitative indicators:

• 54% of backend and 57% of frontend suggestions from AI were accepted by developers;
• 13% of recommendations were postponed due to low priority;
• 100% of teams in the company use the system daily;
• The average PR merge time decreased by ~30%.

Quality results:

• The workload on senior engineers has decreased;
• Improved consistency in code style and quality across teams;
• The developers noted a reduction in "cognitive fatigue" during the review.

Technological cooperation

The project was implemented in partnership with WiseIT and Google Cloud Platform, which provided a stable infrastructure for processing LLM requests and scaling solutions.

Gemini API has become a key component of the architecture — it allows you to create context-dependent queries that "understand" the specifics of our platform's code.

The AI reviewer not only detects syntactic errors, but also evaluates compliance with the business logic of the product, suggesting when changes in the code may disrupt critical connections or processes.

Fintech is one of the most sensitive industries for working with data. That is why Tranzzo's AI solution is deployed within its own GCP environment, without transferring data outside the company.

This ensures:

• complete code privacy;
• control over computing resources;
• PCI DSS compliance.

GCP also provides a transparent cost model: resources for AI workloads are paid for on a pay-as-you-go basis, which reduces financial risks and allows you to scale without excessive costs.

The results of implementing Code Review Automation exceeded expectations:

• 54–57% of developer feedback showed improvement in AI feedback quality after training the system.
• Reducing the time spent on code review has reduced the workload on senior engineers and accelerated releases.
• Reduced risk of critical bugs — thanks to early detection of errors in logic and standards.

For businesses, this means faster time-to-market, greater product stability, and better use of human resources.

AI is no longer an experiment — it has become a tool for improving efficiency.

For us, AI is not a passing trend, but a strategic vector for development. We decided to implement intelligent solutions in our internal processes so that each stage of technology creation would be efficient, stable, and scalable.

The inclusion of the Tranzzo case study in The SMB AI Playbook: Driving Real Value in EMEA confirms that innovation is possible even in highly regulated industries. We will continue to share our experience and discover new ways in which AI can help businesses move faster, smarter, and safer.

🔗 The white paper is available in the section “How businesses are using AI across EMEA today” EMEA SMB AI from Google Cloud.
Learn more about Tranzzo's technology solutions at tranzzo.com.