This week has been a big week for AI companies in enterprise contracts, with Zendesk announcing a new AI agent that it says can solve 80% of customer service problems, Anthropic and IBM announcing a strategic partnership, and Deloitte announcing a deal with Anthropic. Additionally, Google announced a new AI platform for business.
That doesn’t mean it’s all smooth sailing for large companies leveraging AI. In fact, the timing of Deloitte’s announcement is a bit troubling, coming on the same day that Australia’s Department of Employment and Workplace Relations announced that professional services and consulting firms that submitted a number of supposedly hallucinatory AI-generated reports to the department will have to pay refunds.
On the latest episode of the Equity podcast, Kirsten Korosec, Sean O’Kane, and I discuss the latest AI headlines, contrasting them with last week’s news about the new Sora app. While AI companies may ultimately make real money from consumer social networking apps, doing business with companies offers a more direct path to big profits.
You can read a preview of the conversation below, edited for length and clarity.
Anthony: I think this actually ties into what we were discussing last week about some of the GenAI social networks. We saw this as a potential way these AI companies could ultimately make money. I definitely think that’s the case, but we have a long way to go to get there. And while businesses may not find it as interesting or appealing as consumers, that’s actually where the real money is.
Maybe Sora is how OpenAI will make money in 5 years, but this is how these companies will make money now.
And Deloitte (News) was particularly shocking. Just pointing out that these models aren’t (always) perfectly ready for prime time can sometimes feel like a bit of a breaking record, but I think it’s reassuring that the Australian government has actually pushed back and said, “No, we can’t do that.”
tech crunch event
san francisco
|
October 27-29, 2025
I’m not saying that AI should never be used to create these types of reports, but I think you could make that argument. But if you do that, you actually have to take responsibility for the output. You should actually look it up and make sure the information quoted is genuine. You can’t just feed that into a model and say, “Okay, job done. Take any number of billable hours.” I think whoever did that should be ashamed and fined.
Kirsten: Of course. Sean, Zendesk also announced this week that they’re actually creating a tool that handles almost all of customer service and basically takes humans out of the process. Are you starting to see that kind of (automation) starting to permeate your daily life, the way you move around the world, or how automakers approach services?
Sean: Yeah, I’ve actually written about that a few times. There are a number of different startups developing complete customer service suites, voice agents, and LLMs for email and text messages from dealerships and service centers. Actually, I think it’s a worthwhile idea. Because the problem isn’t there. Because there aren’t enough people to do the job, their jobs will be taken away. That means missed calls or bounced calls.
Especially when I go to the service, I get sent to the service department. Everyone is busy. So if we can capture that accurately and make it easier for people to get a response, the question for me is to what extent will companies adopt it and continue to use it? All sorts of technologies have come along over the years, like web forms, and merchants have been doing it, and then they’re forgotten. And it’s just posted on their website so you think it’s going to work, but it doesn’t because they just want you to call them.
So I’m somewhat optimistic and hopeful that something like this might actually be the first touchpoint between people and (businesses). And it looks like it’s about to become clear.
Equity is TechCrunch’s flagship podcast, produced by Theresa Loconsolo and posted every Wednesday and Friday.
Subscribe on Apple Podcasts, Overcast, Spotify, and all casts. You can also follow Equity on X and Threads at @EquityPod.
