This has been a big week for AI companies signing major enterprise deals. Zendesk unveiled new AI agents that are supposed to be able to resolve eighty percent of customer service issues. Anthropic and IBM announced a strategic partnership, and Deloitte also announced a new deal with Anthropic. In addition, Google announced a new AI-for-business platform.
This does not mean it will be smooth sailing for large organizations using AI. The timing of the Deloitte announcement was a bit awkward, as it came on the same day the Australia Department of Employment and Workplace Relations said the professional services firm would have to pay a refund. This was for delivering a report that appeared to contain a number of AI-generated hallucinations.
On the latest episode of the Equity podcast, the hosts discussed the latest AI headlines and contrasted them with last week’s news about the new Sora app. While AI companies may eventually make real money from consumer social networking apps, enterprise deals offer a more immediate path to significant revenue.
A preview of the conversation is included below. One host noted that while consumer applications are often seen as more interesting, the enterprise is actually where the real money is right now. Sora might be how OpenAI makes money five years from now, but these enterprise deals are how companies will make money today.
The Deloitte news was especially striking. It can feel repetitive to point out that these models are not always ready for prime time, but it is encouraging that the Australian government pushed back and said this was unacceptable. The argument is not that no one should ever use AI to create reports, but if you are going to use it, you must be responsible for the outputs. You have to verify that the information is real. You cannot simply feed a task into a model, consider the job done, and bill for the hours. Anyone who does that should be embarrassed and fined.
The conversation then turned to the Zendesk announcement. They are creating tools that will handle nearly all of customer service, essentially removing the human from that process. The question was raised about whether this kind of automation is starting to appear in everyday life, such as how automakers deal with service.
The response was that several startups are developing full customer service suites, including voice agents and language models for emails and texts from dealerships and service centers. This is seen as a worthy idea. The core problem is not a lack of people to do the jobs, but that you can never get someone on the phone or you get bounced around between departments. If this technology can capture issues accurately and make it easier for people to get a response, it could be very useful.
The real question is how much businesses will adopt and stick with the technology. Over the years, other technologies like web forms have been implemented but then forgotten, left to sit on a website without functioning properly because the business ultimately just wants you to call. There is some optimism that this new AI technology could become a customer’s first touch point with a business, and it appears we are about to find out if that holds true.