Table of Contents

I’m as excited as anyone about recent advancements in Artificial Intelligence (AI). Large language models (LLMs) such as OpenAI GPT-4 are especially exciting, as their text generation capabilities are approaching human-level quality. Further, these AI models have proven they can be used to assist humans with a wide variety of tasks, from writing code to analyzing legal agreements. Perhaps most importantly, newer AI models are able to understand Application Programming Interfaces (APIs) and use them to solve problems.

While AI models are getting a lot of attention, API development will continue to be a necessary part of software for the foreseeable future. APIs can enhance the capabilities of AI models, and AI can make APIs more accessible to users. Together, APIs and AI models can enable new business models and solve problems that previously required completely custom automation solutions.

The Age of AI has begun, but that doesn’t mean the time for APIs has passed. In fact, APIs are more important than ever. APIs help us share information and capabilities. AI models can understand APIs and use them to solve problems. Together, APIs and AI models can enable new business models and solve problems that previously required completely custom automation solutions.

What is an API?

Most modern mobile and web applications are built on top of APIs. APIs are an essential building block of modern software development, and are the foundation of the modern internet. With APIs, developers can build applications that leverage the power of other applications and services. This lets apps share data and functionality.

For example, a mobile app might use an API to get the current weather. Your smartphone doesn’t know it’s partly cloudy: that information is coming from a weather service that has made its information available through an API.

In short, APIs are a way to share data and functionality between applications.

Why Do APIs Matter for AI?

AI models are able to understand APIs and use them to solve problems. This is a big deal. It means that AI models can be used to solve problems that would otherwise require a human to write code. Applications that were once disconnected can now be connected through AI models. All that is required is documentation for the APIs, and a sufficiently advanced AI model with Internet access can do the rest.

This is basically how ChatGPT Plugins work: you provide API documentation to the AI model, and it infers how to interact with the API based on users’ requests. While most of the recently hyped AI applications have to do with using capabilities derived from an AI model’s training data, API integration brings new capabilities beyond what the model was trained to do.

New Possibilities

So what do those new capabilities look like? Any service made available through an API can now be utilized by an AI model. That, in turn, means any person who interacts with an AI model can now interact with that service.

Improving the API User Experience

Some APIs are difficult to use. For example, the OpenStreetMaps Overpass API is a powerful tool for querying OpenStreetMaps data, but quickly becomes hard to follow. Utilizing ChatGPT, however, users have built complete queries by simply asking for the result they want.

Enabling New Business Models

With conversational chat bots, some data sources and services may no longer need a custom user interface to be useful to users. Instead, users can simply ask for the information they want, and the AI model can retrieve it for them. All the business owner needs to do is provide API access to the AI model. Certainly there will be opportunities to monetize direct access to APIs through AI chat bots.

Connecting Disparate Systems

AI models can quickly combine data and services from formerly disconnected systems. For example, an AI model could combine data from an issue tracking system and a time tracking system to generate detailed invoices. The AI would need access to both systems’ APIs, some basic context on the relationship of data in the systems. Then a user could simply ask for an invoice, and the AI model would generate it.

The Continued Importance of APIs

While recent AI developments are bringing new capabilities to the table, APIs will remain an essential part of the software ecosystem for years to come. APIs can enhance AI models’ capabilities, and AI models can make APIs more accessible to users. Together, APIs and AI models can enable new business models and solve problems that previously required completely custom automation solutions.

Are you interested in finding how your business can leverage AI with your existing data and services? Contact us to start a conversation about how we can help you build API-powered AI solutions.