How should companies embrace the API-first approach, and why Joyce thinks Apple’s Vision Pro will be a big thing for developers and APIs
Joyce likes to say that now we work and live in times when APIs are eating the universe, paraphrasing Marc Andreessen’s statement from 2011, when he said that software was eating the world. What he meant was that companies that did not have a software strategy would not be able to compete in the world and will eventually fade out.
The same could be now said about APIs. It’s high time for companies to start thinking about their API strategy, or they will not be able to compete.
We’ve recently seen two big social media platforms, Twitter and Reddit, whose only business model for a long time seemed to be only advertising, suddenly realize their APIs could be their product (albeit with some controversies).
Joyce, welcome to Treblle API Talks!
For the first time this year, Postman’s State of APIs Report asked participants if they saw their API as a product, and 60% of them said yes.
What are your thoughts on this shift to API as a product?
API-as-a-Product can be interpreted in two ways. I’ve heard it refers to companies that provide an API as their core offering, such as Twilio, Stripe, and Algolia. Later on, those companies may expand to additional services and interfaces.
Another way to think about API-as-a-Product is within organizations that view an API as a layer on top of a bigger system. For those organizations, an API is another channel to deliver their core product.
More and more organizations are approaching their API with a strategic lens, some even hiring API product managers to optimize customer value with their public APIs.
You could think of Twitter and Reddit in that second category. Artificial intelligence large language models are using these kinds of companies’ free public APIs to ingest large amounts of data to ultimately monetize their businesses. Reddit’s new pricing announcement allows Reddit to increase revenue from AI LLMs and also shut down third-party applications so that more users remain on Reddit’s platform, allowing them to optimize their short-term metrics as they prepare for IPO.
API-first means prioritizing APIs at the beginning of the development process, creating an API strategy and developing APIs before writing other code, not the other way round.
What else would you say is essential for a company to embrace the API-first approach?
I’ve seen different models work successfully at different companies, so it’s not a one-size-fits-all approach.
At a high level, API-first is an approach that helps organizations that manage APIs at scale.
Because API-first is a design-driven approach, you can get input from stakeholders during the design phase of the lifecycle when it’s easier to make modifications, and the cost of rework is the lowest.
There are a lot of ways to accomplish this. For example, you can use mock servers to prototype an early design. You can use an OpenAPI schema or other contract format to gain alignment among stakeholders. You might even use a text document to detail the decisions and rationale in the design process.
The Challenges: Discoverability and Accountability
There’s technical complexity, organizational changes, documentation, security concerns…What are the biggest challenges in adopting an API-first approach and managing APIs at scale?
Companies face different challenges depending on the maturity of their API. At large organizations with many APIs, I see two challenges that have come up consistently over the last year: discoverability and accountability.
Discoverability. Many teams are building out their internal catalogs of APIs. It’s challenging to gain visibility into which APIs simply exist across large organizations, so there ends up being redundancy and wasted effort.
Not having an easy way to discover internal APIs also negatively impacts the ability to reuse schemas, establish consistent design patterns, and enforce governance rules.
Accountability. The person or team tasked with enabling an API-first approach and, ultimately, for enforcing API governance has a big responsibility. Most teams are pursuing a policy-as-code approach to automate feedback loops, but it’s also important to have human feedback loops and clearly communicate API standards.
Documentation, performance and reliability
What are the essentials every engineer has to have in mind when it comes to working with APIs?
When it comes to consuming APIs, factors change in importance throughout the user’s journey. For example, at the beginning of a user’s journey, having adequate documentation and consistent error handling is most important.
When the user begins to deepen their usage, having SDKs and guides for advanced scenarios becomes more important. When the user decides to ramp up their usage, it’s all about API performance and reliability.
This means that API producers must consider the entire user journey when considering the architecture, design, and overall experience to flexibly accommodate the user throughout their own lifecycle.
Apple Vision Pro and APIs
LLMs are already driving a spike in API adoption. What other similar trends do you see as important in the next couple of years?
I usually predict Web3, low-code technology platforms, and AI as the drivers of API growth, but since it was presented, I have to add Apple’s AR headset.
While it’s not even available to purchase, Vision Pro is a big move for developers and APIs.
In the same way, the launch of the Apple App Store initiated a wave of mobile development, there is now a new technology and new paradigms to understand. Apple will incentivize the early generation of development to ensure the ecosystem is properly seeded when the headset finally becomes available to the general public.
API Monitoring is too often an afterthought
What is your approach to API monitoring/observability?
Like quality or security, API observability can best be accomplished when considered early in the API lifecycle. But too often, API monitoring is an afterthought.
And so there’s demand now for technology that does the work for you later in the API lifecycle, such as machine learning to distill insights from vast amounts of data, auto-generating documentation, and tracking deployed endpoints you didn’t even know existed.