BreakpointThe PayPal Developer Blog has a new home. Readers are directed to continue finding developer news, tutorials, product updates, and engineering insights on the new PayPal community blog.
PayPal engineering blog
PayPal announced the Agentic Toolkit, a library that helps integrate PayPal’s orders, invoices, disputes, shipment tracking, transaction search, and subscriptions APIs into AI agent workflows. The toolkit provides pre-built abstractions to simplify API interactions, supports TypeScript (with Python support coming soon), and is compatible with MCP servers and other AI agent frameworks to enable agentic commerce.
PayPal announced availability of its Model Context Protocol (MCP) servers to enable agentic commerce, allowing developers to connect AI agents to PayPal services via an MCP endpoint. The rollout supports both local MCP servers (using the Agentic Toolkit) and remote MCP servers with authentication, beginning with an invoicing feature and with more capabilities to follow.
PayPal Data Science describes Delta CV, a metric for measuring incremental customer profit margin over 12 months after product adoption, and explains how they estimate it using causal inference with synthetic controls. The methodology matches adopters to synthetic controls via KNN on pre-adoption transactional features and discusses interpretation, caveats, and reliability checks for Delta CV estimates.
This post outlines PayPal’s Cosmos.AI platform, an enterprise AI/ML platform designed to provide end-to-end machine learning development lifecycle (MLDLC) capabilities, MLOps, and governed self-service for thousands of users. It describes guiding principles such as responsible AI, abstraction for best-of-breed components, multi-tenancy, and how the platform supports generative AI and large language model workflows.
PayPal describes how they migrated a heavy Apache Spark pipeline to Spark 3 and NVIDIA’s Spark RAPIDS to leverage GPUs, tune partitioning (AQE and spark.sql.files.maxPartitionBytes), and reduce cloud costs. The post walks through tuning, GPU-cluster migration challenges (memory pools, local SSD spill), and final cost/utility comparisons showing up to ~70% cost reduction for eligible workloads.
This post explains PayPal’s declarative (config-based) feature engineering approach to speed feature time-to-market and reduce total cost of ownership. It outlines feature complexity tiers, self-service tooling for feature declaration, time-travel/backfill support, and how automation and reuse improve scale and maintainability.
PayPal introduces a Visual Studio Code extension that helps developers integrate PayPal checkout solutions by providing samples, snippets, quick links, and commands directly within VS Code. The extension aims to reduce integration friction, speed development, and streamline access to PayPal developer resources.
This post explains how merchants can implement recurring payments on mobile web using Apple Pay together with the PayPal JS SDK and V3 Vault. It covers how tokens are stored for future charges, how payment indicators and buyer authentication reduce rejections, and the differences between fixed, variable, and unscheduled recurring payment use cases.
A step-by-step guide to integrating PayPal’s Standard Checkout using the PayPal Buttons component, demonstrating both basic and custom integrations with an example shopping app. The article walks through required prerequisites, front-end and back-end code (HTML, Node.js/Express), cart handling, and customizing the checkout flow, with examples using TailwindCSS.
This post promotes PayPal’s Developer Meetup at Money20/20, outlining reasons developers should attend including networking with PayPal leadership, learning about PayPal Complete Payments (PPCP), and previews of PayPal USD (PYUSD) stablecoin. It also highlights sessions on next-generation experiences powered by AI and other developer-focused topics.