APIs Power Intent-Based Next Generation UIs
AI limits disappear with the help of APIs
APIs are the missing link between AI and intent-based user interfaces. While AI alone has the power to interpret natural language, its hard skills are limited to what it can manipulate. To be an effective user interface, AI needs to be able to integrate with the systems that are part of users' workflows. This is precisely the void APIs are filling.
This article is brought to you with the help of our supporters: Zuplo and Speakeasy.
Zuplo is the only API Gateway that runs your OpenAPI definition directly. If you care about high-quality developer documentation, API analytics, and zero configuration deployments, give Zuplo a try today.
Speakeasy provides you with the tools to craft truly developer-friendly integration experiences for your APIs: idiomatic, strongly typed, lightweight & customizable SDKs in 8+ languages, Terraform providers & always-in-sync docs. Increase API user adoption with friction-free integrations. Try Speakeasy now.
Intent-based outcome specification is a UI paradigm where users convey to the system the outcome they want. Instead of instructing the system with commands of what it should do, users simply present the final outcome of what they need. The concept has been recently described by Jakob Nielsen, who claims that on AI systems "the user tells the computer the desired result but does not specify how this outcome should be accomplished." What happens when the desired result is beyond the capabilities of the AI system? You'll most probably get an answer saying the system can't produce the outcome you desire. And that's precisely the kind of tasks APIs can carry out.
APIs are the engine behind integrated products. The concept of "whole product" in marketing establishes that the needs of customers go beyond what a single product can offer. Regis McKenna, the marketer behind Apple's first computers, defined the concept as a "generic or core product, augmented by everything else a customer needs." In the digital world, you can extrapolate this concept to provide products consisting of features that are integrated with a common user interface. The role of APIs is then to provide the integration mechanisms to link the user interface to the features that constitute the whole product.
Traditionally, APIs have been the glue between different systems. By offering a programmatic way to execute remote functions, APIs let integrators augment products to meet the demands of their customers. Integration tools like Mulesoft and Boomi offer powerful services that let developers integrate different systems to obtain new solutions. Another type of API integration is the one that powers visual UI features. By integrating with third-party APIs such as Stripe and Algolia, developers can implement payment and search capabilities into their applications. In case you haven't noticed, there's a common element in all these integration scenarios. The intent behind the integrations is defined by the integrators—the developers, in most cases. On a conversational interface such as ChatGPT, the scenario is different, though.
When you have an AI-powered conversational interface where you can express your intents, it's only natural that you want to extend its abilities by integrating it with external features. Any non-technical user interacting with the UI should be able to use integrations by identifying the outcomes they're looking for. This is exactly what OpenAI is providing with its recently released Actions feature that lets you connect ChatGPT to external APIs. You can extend the hard skills of the AI system by providing an OpenAPI definition of the API behind the action. Now, the outcome you expect can be fulfilled by an external API whenever the AI system isn't capable of handling it on its own. Existing integrations with external services include the ability to order from grocery stores, search for flights, book a table at a restaurant, and Zapier, opening the door to a wide range of connections.
Zapier is a good example of OpenAI's extensibility in action. The iPaaS company has been at the forefront of extending ChatGPT with the numerous integration capabilities it offers. They first started with their ChatGPT plugin that captures a user's intent and attempts to fulfill it using an existing integration. Since then, they've been adding more AI-powered features and integrations. Another example is Blobr, an AI "copilot" that connects to your APIs and lets users engage conversationally. Blobr's technology lets you connect your OpenAPI definition to an engine that streamlines the process of interacting with AI and establishing the conversational UI. Bricks goes even further and lets you extend your company's internal tools by integrating with different APIs. With Bricks, you can, for instance, retrieve all your sales leads from one app and then create a task on another app to contact each lead. All this orchestration is hidden from users and follows an intent-based approach.
All these solutions have an inherent business risk. By becoming too attached to OpenAI's product offering, these companies become vulnerable to any changes. OpenAI might decide to offer competing features that can take whole companies out of business. Additionally, the cost of using OpenAI's API is not negligible. Overall, from a business perspective, I hope that companies like the ones I mention in this article find an alternative viable solution to using OpenAI exclusively.
This is just the beginning of a new generation of intent-based UIs. Right now, OpenAI is at the center of the innovation. However, competition in the AI space from systems like Bard and Anthropic will open the door to an entirely new range of startups. Also, AI open-source tools and models like LangChain and Hugging Face Transformers are gaining traction and are becoming viable alternatives to commercial solutions. It's only a matter of time before we see APIs powering the next generation of UIs.