Systems | Development | Analytics | API | Testing

As we close out 2025, it's time to reflect on the hard lessons learned from deploying AI agents in production environments. The promise of AI agents is compelling: autonomous systems that can handle complex tasks, make intelligent decisions, and execute actions on our behalf. But as several high-profile incidents this year have starkly demonstrated, this autonomy comes with unprecedented risks when proper identity and access management controls are absent.

API protocols evolve every few years. We have moved from SOAP to REST, then to GraphQL, gRPC, and AsyncAPI for event-driven systems. Now with the rise of large language models (LLMs) and AI agents, organizations need a new class of interfaces that allow agents to take action across real systems, not just generate text. LLMs are powerful reasoning engines, but they lack context. They cannot perform actions by themselves, see real-time data, private information, or internal systems.

If you're working with APIs, chances are that the term "API gateway" has popped up in a conversation or web search at least once. Although they are not new, many developers that are new to creating APIs (or deploying them into a production environment) haven't heard of them or don't fully understand what the technology does. In the most simple terms, an API gateway is a single entry point where API traffic comes in and is then proxied to your actual APIs.

Choreo already provides application developers with powerful CI/CD pipelines, enabling them to automate, customize, and streamline build and deployment processes. However, until now, external resource integration was limited strictly to APIs—lacking a native mechanism to create or register other external dependencies like databases, message queues, or storage services directly within the platform.

AI has become an integral part of how we work today. From writing code to managing workflows, AI assistants help us complete tasks faster and with greater precision. In modern software products, AI copilots are no longer a luxury, they’re a necessity. They reduce cognitive load, simplify complex workflows, and allow users to focus on what truly matters.

Applications and systems using AI have exploded in popularity, with every company looking to integrate AI anywhere they can. This move toward AI-assisted and AI-powered products appears to be the future. However, early adoption is great, but gaps form quickly at scale. For example, in 2023 OWASP began to publish the OWASP Top 10 for LLM Applications (updated again in 2025), which outlined ten common security flaws found in LLM-based applications.

This blog expands on insights Amjadh Ifthikar shared during his WSO2Con Asia 2025 presentation. Watch the full session here.

APIs are no longer just simple integration interfaces. As AI models, agents, and digital ecosystems become increasingly interlinked, APIs have become products. They have their own usage patterns, costs, and business implications. To make informed, data-driven, and business decisions, you require deep visibility into your systems. To address these challenges, the latest v4.6.0 release of WSO2 API Manager includes a dedicated API intelligence engine, powered by Moesif.

WSO2 Integrator, which unites the BI, MI, SI, and ICP components, continues to evolve to help enterprises build and manage integrations that are faster, smarter, and easier to operate. With the November 2025 release, the product becomes more cohesive, intelligent, and developer-friendly than ever before. Included in this release: Together, these updates empower teams to innovate faster, govern integrations effectively, and operate reliably across the enterprise.

AI adoption is accelerating across every industry, from customer service and marketing to finance and retail, and we’re still far from its peak. Many organizations have already begun integrating AI agents into their business operations. A key aspect of this transformation is the use of AI agents, whether purpose-built or off-the-shelf, to assist with or even replace tasks traditionally handled by humans.