AI hasn’t just been trending in 2024. Tech evangelists were already advocating for “AI-first” strategies as far back as 2021, when Ash Fontana published The AI-First Company: How to Compete and Win with Artificial Intelligence. But what does it mean, exactly, to be AI-first?

At the beginning of this year, former Gartner Chief of Research Rajesh Kandaswamy wrote a LinkedIn post, Building an AI-First Company – An Introduction. In it, he lays out the principles of going “AI-first,” an idea that’s been gathering momentum in tech-savvy circles. He defines an AI-first company as prioritizing “the use of AI to accomplish anything, rather than relying on established processes, systems, or tools.” However, AI-first doesn’t mean AI-only, as he’s quick to point out. When AI tools fall short, they fall back on traditional methods, for the time being at least.

What does it mean to be AI-first? What are some scenarios that AI is uniquely equipped to handle? Last but not least, what are these situations where AI stumbles, causing users to fall back on traditional systems, processes, and tools? To find out we’ve delved into AI-first design and principles to find out more about this emerging trend.

Why AI-First?

AI-first companies begin with AI rather than simply augmenting existing practices. Leading with AI informs every business decision, from the website layout to the data gathered. AI-first thinking lets business owners consider these issues ahead of time, giving them an invaluable advantage in today’s AI-driven marketplace. It also offers a unique opportunity for startups, as they’re less entrenched in established ways of thinking and doing business.

Kandaswamy opens his article with thoughts that led him to pursue AI-first as a viable business model. His first belief was that AI stands apart from other technologies and business models. Corporations are merely a collection of human beings, after all. This leads to his first premise, that AI shouldn’t replace humans but be used to further human goals.

According to Kandaswamy, this requires an AI-first approach. Simply integrating AI into existing processes would tether it to limited human thinking. This is also why he branched out to start his own AI-first company, as startups are less beholden to legacy systems and practices. It’s also why he’s starting now, as AI will only get faster, cheaper, and more effective.

The Benefits of AI-First

Creating an AI-first business gets business owners in the habit of thinking about AI from the outset. One of the first and biggest benefits of AI-first is making the most of your data. The hope is that AI-first also allows businesses to be faster, leaner, and more efficient. It can streamline everything from code generation and testing to generating sales and marketing materials. It can also optimize customer service by using sophisticated data analysis to deliver superior personalized services. AI-first also allows business owners to identify and enable novel products and services. Finally, it further encourages optimization and efficiency by employing sophisticated data analysis and eliminating redundant, time-consuming tasks.

In an interview with TechTarget from 2021, Ash Fontana, author of The API-First Company, references Google as “the original AI-first company,” citing their emphasis on collecting valuable user data to power machine learning to optimize their products. This is an excellent example of how user data can be leveraged to drive new products and services and generate novel income streams. Think of the sponsored posts and suggested selling that dominate the top of search engine results pages (SERPs) for one concrete example of AI-first principles in action. Google is a prime example of how gathering the correct data and using it intelligently can become a massive source of revenue when handled correctly.

A recent post from Infosys on Financial Times illustrates how an AI-first approach allows them to streamline their operations. They’re already using AI to write and maintain code that is 80% more efficient. They also use AI to monitor financial transactions, significantly reducing the amount of fraud, which is another area where AI excels. They even use it to analyze mundane tasks like supply chains and utility usage, allowing them to further reduce expenses.

AI-First In Action

As we have briefly seen from these above examples, AI-first isn’t just some abstract design philosophy. In fact, it’s already being implemented in practical, concrete ways that generate major income and deliver actual savings. These are just a few of the countless examples of how AI is already being used. Yum Brand, the parent company that owns Taco Bell and Pizza Hut, is employing AI-first principles to streamline everything from scheduling to prep work in kitchens. Forbes was calling Uber AI-first back in 2018, thanks to their widespread adoption of AI for a range of tasks.

In a different edition of his AI-first newsletter, Kandaswamy gives some real-world examples of AI-first. The first practical example he puts forward is app redesign. In this example, he mentions his partner using GitHub CoPilot to suggest a new and novel app architecture. He then used GitHub CoPilot and Visual Studio Code to break the project into small, manageable steps. He then used the same process to create the modular blocks of code.

The second example of AI-first in action illustrates using AI to create financial strategies. To start, he used ChatGPT to weigh the pros and cons of different investment scenarios. He got the best results while using ChatGPT to simplify advice he’d received about SAFE. Here’s where some of AI-first’s limitations appear. We’re a long way off from simply automating an investment strategy. However, AI can still be useful in clarifying complex login into simple, streamlined thinking.

Thirdly and finally, Kandaswamy uses AI for testing resources, which is a widespread application. He had the best luck using AI to generate a checklist of things that needed to be tested, which generated many suggestions he hadn’t thought of. In future experiments, Kandaswamy plans to use AI to create automated testing.

The Future of AI-First Design

AI isn’t going anywhere. It will only keep getting faster, cheaper, and more capable. As such, AI-first design will only keep getting more prevalent. We’re still a long way off from AI being able to start and run businesses all on their own, but that day could come. Even today’s AI is capable of streamlining operations, clarifying complex instructions or data, and generating code. AI is already capable of improving business in a variety of ways, and it’s only going to become more powerful and profitable. Now’s the time to start experimenting with AI-first design.