# How Mastercard Uses AI to Scale API Quality, Discoverability, and Integration
source: https://developer.mastercard.com/blog/how-mastercard-uses-ai-to-scale-api-quality-and-discoverability/index.md

Authors: David Rooney, Shimona Rudra. APIs are foundational to modern payments, enabling everything from checkout and identity to fraud prevention and tokenization. The growth of the Mastercard Developers API platform has driven a significant increase in both the number and complexity of APIs that developers depend on. Ensuring consistent quality at this scale has driven the ongoing evolution of Mastercard's API Gold Standard. The API Gold Standard was originally established to ensure APIs were accurate, consistent, and easy for human developers to understand and integrate. Early assessments focused on high quality OpenAPI specifications, clear documentation, predictable behavior, and alignment with platform conventions, delivered through expert‑led reviews that emphasized rigor and developer usability. Evolving Standards Today, accuracy remains essential, but it is no longer sufficient on its own. APIs must also be easy to discover, understand, and integrate when consumed not only by developers, but also by large language models (LLMs), copilots, and autonomous agents. Developers increasingly rely on Integrated Development Environments (IDEs), AI assistants, and conversational interfaces rather than traditional keyword search, and generative systems often consume API specifications and documentation directly. To support this shift, Mastercard emphasizes Generative Engine Optimization (GEO), structuring API content so generative systems can accurately retrieve, interpret, and recommend the right APIs even when prompts do not exactly match naming. You can see below where Mastercard Digital Enablement Service (MDES) surfaces when a developer prompts "How do I tokenize a card?". To meet these requirements at scale, Mastercard has built AI‑powered API assessment tools on Databricks, combining automation with expert human oversight to improve quality, consistency, and discoverability across its API ecosystem. Historically, API evaluations examined the completeness and accuracy of OpenAPI specifications, the clarity and structure of documentation, alignment with platform standards, and overall developer usability. While this approach delivered strong results, it became increasingly difficult to apply consistently as the API portfolio expanded. AI‑assisted assessment now enables Mastercard to evaluate API specifications and documentation at scale. Automated analysis identifies issues such as missing metadata, unclear or inconsistent descriptions, structural gaps, and insufficient examples. The output is not only diagnostic, but provides actionable, standard-aligned feedback that teams can apply earlier in the lifecycle. This leads to faster feedback loops, fewer late‑stage surprises, and a more predictable integration experience for developers. How the AI-Assisted Review Works Each assessment begins with automated analysis of OpenAPI specifications and supporting documentation using structured rule sets and large language models. A Databricks‑powered workflow transforms raw Markdown documentation into structured, evidence‑backed inputs. Relevant sections are identified and evaluated against a consistent set of standard-aligned criteria. The workflow generates a comprehensive assessment report, using Databricks Model Serving to perform the LLM-based analysis, while findings, configurations, and reports are stored in Unity Catalog. This creates a reliable and repeatable starting point for reviewers, ensures consistency across assessments, and significantly reduces manual effort. For developers, the result is clearer documentation, better examples, and APIs that are easier to discover and integrate. Why Human Review Remains Essential Automation alone is not sufficient, particularly in a regulated industry like financial services. Every AI‑generated assessment is reviewed by experienced API and standards experts who validate findings, apply product and domain context, and make final determinations on readiness and quality. This human‑in‑the‑loop approach ensures AI accelerates insight without becoming the final authority. It enables nuanced decisions, handles edge cases correctly, and keeps feedback practical, accountable, and grounded in real developer needs. Human review is especially important as APIs are increasingly consumed by copilots and autonomous systems. Reviewers ensure API designs support these usage patterns through high-signal metadata, consistent terminology, documentation structured for reliable retrieval, clear mappings to real-world use cases, and unambiguous examples, FAQs, and error guidance. What This Means for Developers For developers integrating with Mastercard APIs, this approach delivers tangible benefits: Spend less time searching and more time building with APIs that surface predictably across search and AI‑powered tools Reduce integration risk and rework through consistent documentation, examples, and behavior Adopt next‑generation tooling with confidence, knowing APIs are designed for AI‑assisted and automated development Building for the Next Phase of API Consumption The API ecosystem is entering a new phase. By combining AI‑powered assessment, expert human review, and a deliberate focus on discoverability and structured content, Mastercard is building APIs that scale with demand, adapt to new consumption models, and remain trustworthy in an increasingly automated world. As AI continues to reshape how software is built, Mastercard's API ecosystem is designed to evolve alongside it, supporting both developers and the intelligent systems that now build on their behalf.