What Does Being AI-Native Really Mean?

The software solutions market has seen over the past year a proliferation of companies and platforms that claim to be AI-native.

Jeffrey Bussgang, a senior lecturer at Harvard Business School and co-founder and general partner at Flybridge Capital Partners, defines AI-native companies as the sum of their parts, particularly employees.

AI-native employees “are wildly adept at using a wide range of modern AI tools in their Jobs To Be Done quest – a skill acquired through intentional and frequent experimentation. This approach allows them to be 10x more productive than the typical [employee],” he says in this LinkedIn post.

Meanwhile, Bussgang defines AI-native companies as being organizations “made up of a collection of AI-native employees who infuse AI into everything they do — every function, every process, and every role.”

As with much of what comes out of Harvard, the definition is a bit ridiculous when faced with some common sense. Only an absolute purist would demand that every process and every function in a company from the CEO to the janitor be adept at and indeed use AI for everything that they do. 

Indeed, according to Bussgang’s definition, only a fraction of today’s startups, which are a small, small fraction of the world’s total companies would possibly qualify as AI-native. Consequently, what meaning or value does the term denote that makes it useful except for hyping a certain type of investment opportunity? It wouldn’t matter to the vast majority of us, unless we worked at venture capital firms or wanted to become angel investors.

What Part(s) of the Company Need to Have AI?

Does it matter if the entire sales team is using AI to try to 10x their conversations with prospects to close more deals versus having an old school high-touch white glove approach that leads to just a few massive enterprise deals? If the numbers are the same or better using the old school approach, who cares that the sales team used AI? More importantly, why does it matter for the end user of the company’s product?

I would strongly argue that the only judge of whether a company is AI-native or not is its products. In this respect, I agree with and will borrow the definition used by the IBM blog,

“AI native refers to something—usually a product, company or workflow— that was designed from the ground up with AI as a core component, not bolted on later as a mere feature.”

Being AI-native in Fraud Prevention

Older fraud platforms were built around static rule sets and later added a machine learning layer to catch what the rules missed. AI-native companies skip that entirely and treat the statistical model as the primary decision-maker from day one, which usually makes them more adaptive but sometimes less interpretable without deliberate, additional effort.

For some legacy anti-fraud vendors, becoming AI-native means rebuilding their solutions from the ground up. 

There are even companies specializing in creating AI agents for existing fraud platforms. However, to paraphrase, Alisdair Faulkner, CEO of Darwinium, an agentic security and fraud prevention platform, when you “agentify” anything, it’s only as good as the underlying application. The agents just make processes more efficient, but if the underlying application is not built to tackle the current set of challenges facing the market, no amount of agentification will change that.

Meanwhile, for many of the startups founded in recent years, being AI-native just means doubling down on what they were already doing – incorporating AI into core products and increasing efficiencies by leveraging more AI technology in workflows. This often translates into adding LLM based technology for parts of the solution that interact with humans, while continuing to do a lot of heavy analytical work with other forms of machine learning that are better suited to analyzing structured data at scale at a cheaper cost.