AI and APIs: Solving Governance and Complexity at Apidays Paris

As organisations adopt more APIs, their ecosystems become increasingly tangled and unwieldy. Multiple speakers touched on the high costs and delays associated with API sprawl, with more robust governance emerging as a key aspect—and AI as an unexpected assistant.

Perhaps AI might finally help organisations move beyond managing messy APIs to optimise their ecosystems for real growth.

The complexity isn’t just seen within organisations, but also in the proliferation of API solutions. Providers are sprouting like mushrooms, as evidenced by the increasingly crowded API Landscape board.

This mirrors the state of the Open Banking industry, where various open banking API solutions have rapidly emerged in recent years. This surge can definitely offer more variation and choice, though it could trigger a form of choice overload.

Navigating a saturated market and ensuring interoperability between many solutions is rarely a smooth process.

There was an intriguing emphasis on the alleged symbiotic relationship between AI and APIs. AI depends on APIs for data access and functionality, while APIs will increasingly rely on AI to simplify interactions and handle their chaotic growth.

For example, we introduced Opey during the event, which is a conversational tool to simplify API discovery for external and internal teams. It helps navigate vast API catalogues and suggests the right combinations of APIs for specific use cases.

This feels like either a fast-growing trend or simply the logical next step. Postman launched Postbot to help teams debug APIs and write tests faster,  while Axway continues to emphasise AI’s impact on API management.

At the same time, new tools such as TaskMatrix.AI and AnyTool demonstrate how deeply intertwined AI and APIs have become. These systems would rely on thousands (even millions) of APIs to access functionality and data, but it’s AI that makes this scale and complexity possible. In other words, the two are feeding into each other.

This mutual dependence could change how we approach both technologies. APIs will need stronger governance, while their documentation must become more consistent and machine-readable to provide AI with a format it can readily digest.

We also showcased Opey II, an agentic version that shifts the focus from guidance to direct action.

The term “agentic” refers to AI systems capable of autonomously taking actions and dynamically refining their approach to achieve certain goals. In Opey II’s case, these actions are executed through APIs.

For example, a Head of Digital or Product Manager might ask “What were our least used APIs over the past six months?” to find opportunities for optimisation. Opey II adheres to the user’s roles to decide and securely call the relevant APIs, providing insights and helping non-technical roles engage with data through a conversational interface.

There are other reasons why AI agents are all the rage. A different speaker referenced the strawberry problem (LLMs have difficulty counting the r’s in strawberry) to illustrate how agentic AI can overcome certain limitations. 

In short, GPTs don’t “see” text like humans do, they process words as patterns, which can lead to errors if counting specific letters. Agents can refine their approach by generating and running code to find the correct answer.

Axway’s contribution highlighted the critical need for better governance and management of APIs. As organisations adopt more APIs, unmanaged and “shadow” APIs pose security risks. 

They shared some curious statistics from their State of Enterprise API Maturity report:

• 78% of study respondents didn’t know how many APIs their organisations had.
• 98% of organisations find measuring API metrics difficult – despite metrics being foundational to an API strategy.

There was also emphasis on:

• The value of a centralised API catalog to track assets.
• A growing trend of treating APIs as products with the intent to monetise.
• The importance of aligning IT and business goals.

These points resonate strongly, yet it’s interesting how familiar they feel. Such discussions have echoed in conference rooms for years. Is this recurrence due to us circling the same old challenges, or is it becoming increasingly important to highlight these gaps?

One of the key highlights for the OBP team was meeting up with Ismail Chaib, a former colleague and a driving force behind the Open Bank Project in its early days.

It’s easy to forget that attending Apidays also means reconnecting with old friends and partners, which truly completes the experience.

The event delivered a mix of inspiration and familiar challenges. While the AI buzz occasionally felt a little emphasised, it’s clear that the relationship between AI and APIs is evolving into something special—a dynamic we’re all trying to figure out in real-time.

This also stresses the importance of refining tools like Opey II—a more polished version of which we’ll be demoing very soon.