Software? No way. We’re an AI company now

As their stocks tank, software makers are rebranding themselves as AI innovators. Sparkle emojis are everywhere, and some efforts have been more successful than others

When ChatGPT came out in November 2022, Eoghan McCabe was about a month into his second stint as CEO of Intercom, the customer service software company he had co-founded a decade earlier. Growth had ground to a near halt, and now he had another problem.

McCabe foresaw an existential crisis: If chatbots could answer customer queries, and if they replaced customer service agents, companies “wouldn’t need help desks — they wouldn’t need the software anymore.”

He decided that Intercom needed to become an artificial intelligence company, or risk becoming obsolete.

More than three years later, similar realisations about the potential for AI to disrupt software are suddenly rippling through public markets and wreaking havoc on
the industry.

And that existential crisis has spawned an identity crisis: Software companies at every stage — from startups to publicly traded corporations — have been working hard to be seen as AI companies. Sparkles emoji and magic wand icons are everywhere, “.ai” has become a popular website domain, and venture capital investors say almost all of the pitches they’re fielding from software startups have some kind of AI angle.

The reason for the panic is clear: Over the past 12 months, software stocks have seen their biggest plunge in more than 30 years, wiping out $2 trillion of market capitalisation from the peak, according to J P Morgan. The S&P North American Expanded Technology Software Index has dropped about 20% in just the past month. And shares of software giants Salesforce and ServiceNow are down more than 40% over the past year.

A lot of that damage has come just in the past several weeks, as the release of powerful new AI tools, like Anthropic’s Claude Opus 4.6, have put into sharp relief how quickly the technology is advancing. Much the way that software has long disrupted whole industries, AI’s arrival is redefining what it means to be an innovative tech company.

Pressure building in the startup world

As recently as 2024, PitchBook called the business model known as “software as a service,” or “SaaS” — charging subscription fees for cloud software use — “the bread and butter of venture.”

Now, what energy was left in the fading category has been sucked up by AI, a technology that could change both how software is built and how companies pay for it.

Although a slice of startups were pitching themselves as AI companies a few years ago, today “it’s like almost all of them are,” said Tony Wang, managing partner of 500 Global, which runs startup accelerators and venture funds.

Almost half of venture capital last year went to AI startups, according to CB Insights. And at least 135 of the 151 companies in the latest cohort of startups in the tech accelerator Y Combinator identify as AI companies, according to its directory, including 14 with names that end in “AI”

McCabe successfully made the pivot: Within three months after the release of ChatGPT, Intercom, which remains privately held, released an AI customer service agent. The company’s growth rate soared, and annual recurring revenue ratcheted up to nearly $100 million, McCabe said.

But many companies are still scrambling to make the transition.

For almost a decade, Amelia Lerutte has worked on marketing for SaaStr, a company that hosts events for enterprise software founders. She said the tone among software founders working to build AI into their products had changed from just a year ago.

“I think now there’s definitely a bit of panic around it — I don’t want to say panic in a bad way, but just like they’re trying to catch up,” she said.

To reflect the focus of the founders in its community, SaaStr has changed its name to SaaStr AI. And Lerutte’s job title changed to chief AI officer. She now spends much of her time deploying AI agents, some of which are open for SaaStr’s community of founders to use.

Last weekend, she used AI coding tools to create a product management system. “It’s pretty crazy that I could do that,” she said. “This is literally a software tool we were paying for last year.”

Omari Rigg, an entrepreneur who mentors startups through the tech accelerator Techstars, said he often saw shallow attempts at AI transformation.

“They’ll change their domain name, for example, from word.com to word.ai, and they will change their entire messaging on their landing page and their LinkedIn to, like, ‘We are an AI-enabled tool now,’ right?” Rigg said. “But other than that, nothing they’re really doing is actually leveraging a ton of actual AI.” Does that strategy work? Largely no, venture capital investors say. “It’s very quick to see,” said Byron Deeter, a partner at Bessemer Venture Partners who has focused on cloud software companies.

AI is disrupting AI

So-called AI wrappers — companies that build an interface that gives users access to the capabilities of another company’s LLM — have already been disrupted. For example, Jasper, then a writing tool for marketers powered by OpenAI, raised $125 million at a $1.5 billion valuation in 2022. But it started to cut the internal value of its shares after OpenAI released ChatGPT and users began going straight to the source for help with their writing, according to The Intercept. (Loreal Lynch, the company’s chief marketing officer, said in a statement to The New York Times that it had since expanded its product to run entire marketing programmes and had seen “2X year-over-year revenue growth.”)

Other categories of AI companies have gone through an entire startup lifecycle so quickly that they’re no longer appealing to early-stage investors. “There are even AI companies that we were interested in maybe a year and a half ago that we’re no longer interested in,” Wang of 500 Global said. “And that is a fast-moving cycle.”

Many analysts and investors believe that public market investors have jumped the gun on writing the obituary for software companies. “The market is pricing in worst-case AI disruption ​scenarios that are unlikely to materialise over the next three to six months,” J P Morgan strategists wrote in a note this month.

Bessemer’s Deeter said that public market investors hadn’t necessarily done the work to separate the software that was meaningfully leveraging AI from what was not actually adding value, and that software would be necessary to monetise AI models.

“Remember, the foundation models only get paid if there are consumers or apps pulling in payment,” he said. “They’re not a business in and of themselves.”

Even with generative coding, said Wang, “it’s building the UI, it’s building the back end, it’s deploying in the cloud, all that I consider software.”