Blog

Why APIs Are The Lifeline For AI

February 14, 2024

Mieke Houbrechts

AI is revolutionizing the way we analyze data. Learn why APIs will be the true driver of AI innovation in 2024.

In 2023, the buzz around ChatGPT put generative AI on the map for good. In the past 12 months, over ten thousand AI-powered apps have been launched as a result. Whatever the challenge, there’s an AI for that.

It’s no surprise though. According to the State of API, 37% of developers mentioned AI-powered apps as the top project they were excited about in 2023.

But is generative AI the one and only driver of this innovation wave in SaaS? We believe there’s more behind this surge of new AI-powered apps. The silent driver behind this revolution is something that’s been around for much longer: APIs.

What are APIs?

If you need a quick refresher, APIs (Application Programming Interface) are the building blocks of modern software apps. Similar to an electrical power adapter, an API lets two software applications or components communicate with each other and exchange data, even if they are fundamentally different systems.

With APIs, you can easily add new components, features or services to an application to build bigger, more complex software products. Think of it like building a housing community: you would build the plumbing and electricity network first, and the houses later. Similarly, with an API-first approach, you first define how parts of your application interact with each other in the API, before building a user interface on top of it.

Schema explaining how APIs work

Why are APIs so important for building AI-powered apps?

APIs open up the potential to inject powerful generative AI capabilities into your software apps quickly. It’s like plugging a new plumbing line into your existing network. Just like the new plumbing line will connect to every house in the community from the get-go, an API can rapidly scale AI across your entire application.

For example, let’s say you’ve built an HR platform that handles recruiting, payroll, benefits and performance management. You’ve built a well-designed API, so your users can do things like:

  • pulling recruiting data directly from LinkedIn
  • sending payroll data to their financial software
  • monitoring key HR metrics on a dashboard you’ve embedded using third-party software

Now, you want to add an “Ask AI” prompt all across your application, so your product users can ask questions like:

  • Show me the top 5 candidates for job position X on LinkedIn
  • Show me all overdue performance management actions
  • Show me a dashboard with the salary distribution for department Y

With an API-first approach, you can hook up any large language model or generative AI service to power that prompt, and it will automatically pull data from all the components in your app. It’s just one more component that connects to the ecosystem you’ve already built. Et voila, with a bit of effort, your app is AI-powered.

The alternative is to train and integrate your own AI model from scratch, which can take months or even years. Or you can choose to rely on any AI features your third-party integrations offer, like LinkedIn or your embedded analytics tool. However, using multiple independent AI capabilities can get messy very quickly, the models can get outdated, and it’s completely outside of your control.

Instead, SaaS companies are better off directly connecting LLMs like GPT-3.5, GPT-4, Mistral, Gemini, Claude or others via API. You’ll be able to add on new AI services quickly, without being reliant on other vendors to keep their AI features up to the latest standard.

In a near-to-distant future, where AI agents that act and think independently become reality, Karel Callens, CEO of Luzmo, believes APIs will come to play an even more important role. 

“AI is evolving at an incredible speed. The idea of an AI agent that can automatically launch a product for you seems distant, but very real. And when it happens, APIs will be the main orchestrators behind it. API-first platforms, like Luzmo for dashboards, Stripe for payments, or Twilio for communication, will become the core building blocks used to launch a product automatically through the power of AI and APIs.”

Why build AI-powered SaaS apps?

Generative AI’s major strength is to speed up repetitive tasks and add efficiency to your processes. So logically, many SaaS and productivity apps are now looking towards AI to improve their user experience. At Luzmo, for example, we use AI-powered apps daily for all sorts of operational tasks:

  • Creating outlines for blogposts or email templates with ChatGPT
  • Generating SQL queries in ChatGPT
  • Creating AI-powered imagery with DALL-E or Canva
  • Cleaning up audio in Adobe
  • Writing unit tests and other coding tasks with GitHub Copilot

With a powerful API, it is much more feasible to add similar AI features to your apps quickly. But just because you can, does it mean you should?

Haroen Vermylen, CTO of Luzmo, has an interesting take:

“If you’re going to drop everything and prioritize AI over other important features on your SaaS roadmap, you might come to regret it. The truth is that your end users don’t care about AI. They care about getting the quickest solution to their problem, and the fastest way to get the job done. AI could be the best solution in many cases, but it’s essential to keep putting your customer’s needs first.”  

You don’t have to drop everything today to start on AI. But if you’re not consciously thinking about how it can contribute to a better product for your customers, there might be some risk lurking around the corner. You might be overlooking opportunities that your competitors will gladly snatch. A well-researched roadmap and balanced product exploration will help you focus on the right priorities.

How AI and APIs will change customer-facing analytics

As experts in embedded analytics, many of our SaaS customers have asked us how AI will impact business intelligence. How will it change the way SaaS businesses offer reporting features to their customers?

We predict that AI will make in-app analytics more user-friendly in two ways: either via native capabilities in BI apps, or within a broader API ecosystem.

AI-powered analytics will be API-first

In an ideal world, SaaS apps will have a separate layer for AI and data visualization. APIs, however, will be the key to connecting the system together. Our CTO, Haroen Vermylen, shares how this would work with Luzmo as an embedded analytics layer.

“You want to avoid that your SaaS platform has a prompt at the top of your app’s homepage, another prompt on your analytics dashboard, and the two of them don’t work together at all. Instead, SaaS businesses can work with Luzmo’s API as an API-first product, and connect it directly to any AI model. This could be an LLM model, or outputs of unstructured text analysis, that are converted into quantitative datasets.”

AI models change every day, and the norm of today might be outdated tomorrow. With an API-first approach, you’re no longer reliant on third-party vendors, and you could start using a new model as soon as the day it’s released.

“You wouldn’t have to wait for Luzmo or another BI tool to integrate support for the future models like GPT-5 or Gemini Ultra, but instead, you can directly swap out the AI model, connect it to Luzmo’s API, and build dashboards programmatically based on an AI prompt.”
Diagram presenting how to integrate new AI models via API

In addition to this flexibility, an API-first approach also allows for richer insights and trend analysis. Haroen explains the vision for Luzmo:

“We are working on giving our users the option to give an AI model access to your data via API. That way, you can ask it to identify trends and patterns in the information. The model can suggest which data it wants to see, which you can then return back into the model, creating a powerful interaction loop that will lead to deeper insights much quicker."

Native AI capabilities for a better BI experience

Although the API approach is extremely powerful, it does require more advanced coding skills. Not everyone has the time, means and resources to build these integrations. In response to this lack of resources, many embedded BI solutions, including Luzmo, are adding out-of-the-box AI capabilities to their UI.

With the main goal of improving the user experience, native AI features in Luzmo will directly help users get insights from their data faster. Our AI Chart Generator, for example, lets users create charts through a simple prompt. Users can either type their own prompt or choose from auto-suggested ideas to explore their data.

Example of AI-powered chart suggestions in Luzmo's embedded analytics platform
Suggested charts based on your dataset

For SaaS product managers building embedded reporting, this considerably speeds up the time to create a dashboard. But also for end-users of these SaaS apps, AI will bring more efficiency in analyzing data, explains Haroen:

“The AI Chart Generator is readily available in our dashboard editor, and we envision making it available for end-users to use as well, inside any SaaS app embedding Luzmo. End-users will have the freedom to ask questions with a prompt, and like that, they can adapt any charts available in your embedded dashboard to fit their reporting needs.”

In the future, Luzmo plans to explore more ways to make the user experience more intuitive with AI, thus Haroen:

“We don’t expect people will necessarily want to use prompts. In the future, we foresee other ways of presenting insights, whether it’s through suggested charts, or an option to add the next best chart based on the contents of your existing dashboard - available to add to your dashboard with a simple tap.”

The future of AI is API

Exciting times are ahead for developers who are hungry to add generative AI to their apps. But it can be overwhelming too. However, with the power of APIs, any app builder can gain momentum quickly! Simply hook up any AI model directly to your SaaS app, and watch the magic happen.

And that includes reporting and business intelligence features too! If you’re looking to spice up your analytical features with AI, Luzmo can help. Whichever route you want to take, we offer native AI capabilities, as well as a robust API to connect your LLM of choice to our powerful analytics layer.

Start a free trial of Luzmo to test our API and AI features, or book a demo with our product experts for a demo today!

Build your first embedded dashboard in less than 15 min

Experience the power of Luzmo. Talk to our product experts for a guided demo  or get your hands dirty with a free 10-day trial.

Dashboard