|
Welcome back! Microsoft, Salesforce and numerous startups have talked up the idea of an artificial intelligence–powered coach that gives feedback to salespeople about their pitches to customers. Databricks is one of the few companies to actually use such a product at scale. Earlier this year, the data-analysis software provider gave 1,000 employees in its sales group access to an AI sales coach from Yoodli, a small startup. Rochana Golani, Databricks’ vice president of learning and enablement, said it would be too costly to give staff individual coaching lessons. Salespeople have the option of listening to recordings of other people’s calls, but they aren’t getting personalized feedback, she said. Yoodli’s software lets salespeople role-play conversations with customers and offers tips afterward on how they can improve. For example, Yoodli could suggest a different way to describe a product, depending on what kind of customer is getting the pitch. However, Golani said Databricks hasn’t yet done any studies to measure whether the technology is helping salespeople close more deals. Yoodli, founded in 2021, charges up to $20 per user per month for individuals and more for large enterprises that need extra features like added security. That pricing suggests Databricks, which generated $2.6 billion in revenue in its last fiscal year, could be on pace to spend more than $200,000 annually on the tool. Yoodli’s CEO, Varun Puri, compares the product to a “batting cage” that lets salespeople practice before going into real-world situations. Google, Korn Ferry, Snowflake and RingCentral are also paying customers, he said. The market for AI sales coaching could soon grow. Salesforce has said it will release such a product, as have smaller sales-software vendors such as Gong. (As Aaron reported Monday, Microsoft has a lot of OpenAI-powered bots it is using internally with an eye toward releasing them publicly.) But the technology isn’t limited to sales. Golani said she used Yoodli to role-play a conversation in which she practiced giving feedback to one of her employees. To her surprise, the AI bot told her she could improve on how empathetic her remarks sounded—feedback a human might have been more reluctant to deliver, she said. Golani said Databricks could eventually make Yoodli available to employees outside sales. Like other companies that offer voice AI products, Yoodli relies on a workaround to make responses faster. Its tool uses text models from Anthropic, Google and OpenAI to create responses before converting the text to speech, instead of using a model that only handles speech. Yoodli this week raised $13.7 million in Series A funding in a round led by Neotribe Ventures, which hasn’t been reported previously. Puri declined to disclose the company’s valuation. Yoodli faces competition from ChatGPT. Given that so many businesses already pay for OpenAI’s software, instead of buying a specialized tool from Yoodli, they could prompt their sellers to practice conversations with the ChatGPT app. But Puri said Yoodli is trying to make its product more appealing to sales professionals by creating drills to help them practice. “We are trying to make practice really fun,” he said. Corporate executives are planning on hefty budget increases this year for AI agents that can automate work tasks, a May survey from consulting firm PwC found, citing responses from 308 executives. Nearly half said their companies were increasing AI budgets by 10% to 25%, and around a quarter planned on greater increases. Just 12% said they weren’t planning on any increase, according to the survey. Unsurprisingly, the function that adopted AI agents most was customer support, with 57% of businesses using agents in that area already or planning to do so. The categories of sales and marketing and of IT and cybersecurity followed closely behind, at 54% and 53%, respectively. The survey shows there are still plenty of AI skeptics. Roughly a fifth of respondents said their companies weren’t using AI agents at all. Dan Priest, PwC’s U.S. chief AI officer, said those executives are adopting a “wait and see” strategy on AI. “They’re hearing a lot said, but they haven’t seen the [return on investment] yet,” Priest said. Microsoft is betting that the future of agents involves the Model Context Protocol, open-source software designed to make agents more useful by letting them tap into different apps. At its annual developer conference this week, the enterprise software giant said it was launching software to let agents interact with services like coding app GitHub and enterprise resource planning app Dynamics 365. For example, a developer could program a chatbot to pull information on a company’s sales leads from Dynamics 365’s customer relationship management system. Microsoft and GitHub are also joining the MCP Steering Committee, a group of companies that will help manage adoption of the open-source software. They are the only members of that committee that have been announced aside from Anthropic, which developed MCP. Still, other big tech companies have united behind the software in recent months.
| 
