Why a small team with AI cannot replace legacy enterprise tech

This article summarizes an episode of a16z’s video series featuring a16z partner Alex Rampell and Atlassian co-founder and CEO Mike Cannon-Brookes.

Image credit: Timmy Loen

The recent drop in software stock prices is a warning sign about AI. Alex Rampell, partner at Andreessen Horowitz (a16z), argues that investors are misjudging which companies will survive the AI revolution.

In turn, this reality check forces business leaders to ask a tough question: is our product easily replaceable by AI, or does it hold lasting value?

Rethinking how software is priced

To understand this risk, leaders must first examine how they make money. Charging for software based on the number of employees using it to do specific tasks can create a major financial vulnerability.

When new AI can easily do the same work as human workers, companies suddenly have no reason to pay for those extra user accounts. This means software providers that charge for easily automated tasks may see their revenue disappear unless they rethink their pricing models.

Rampell begins by pointing out the companies most at risk from AI. He explains, “Category one is like [this]: you have user accounts, the accounts are being used to produce some element of work, but now, you don’t need the accounts anymore to produce the element of work.”

This problem threatens customer service platforms. Rampell notes that a company like Zendesk could be wiped out if AI handles customer questions. In that scenario, he warns, “that income is going to disappear entirely” unless the company changes how it makes money.

Building hard-to-replace systems

However, assuming all software that charges per user will fail is a mistake. While software used for simple tasks is at risk, systems that store a company’s official records are safer. They charge based on the total number of employees just as a way to set a price, not because every employee is doing a specific task in the software.

As a result, because these platforms run the core operations and store the most important data, their position is strong and can be challenging to replace.

For example, Rampell points to human resources software to show why this is true. He notes, “Workday has this great pricing model where… you’re General Electric; you have 340,000 employees. I’m going to charge you per employee per month.”

He adds that because those employees are not using the software to do a specific task, the income is safe.

This also creates a new opportunity: storing the main company data creates an opportunity. Rampell argues that when new AI is used, the future financial value of the companies that store the correct data is “going to go up a lot.”

The hidden difficulty of replacing old software

Programmers often assume that a small team equipped with AI could easily build a modern replacement for older enterprise systems. But this view overlooks the decades of specialized experience embedded in legacy software.

Getting employees to actually use AI