Palona AI gets $10M to build more reliable customer service chatbots that don’t make mistakes – News

A trio of artificial intelligence engineers who previously led projects at Google LLC, Meta Platforms Inc. and Samsung Electronics Co. Ltd. say they have raised $10 million for a new startup that aims to make AI-powered customer sales chatbots more reliable and “emotionally attuned.”

Palona AI, officially incorporated as Proactive AI Lab Inc., said today’s round was led by UpHonest Capital, with participation from Fusion Fund, NEO Investment Partners, plus Maynard Webb and other angel investors.

The round coincides with the official launch of its sales models, which are already being used by customers such as the home security camera maker Wyze Labs Inc. and the wellness studio Mindzero Inc.

Customer service chatbots have become all the rage ever since the emergence of large language models that power applications like OpenAI’s ChatGPT. Brands are leveraging chatbots as a way to ease the load on their customer service teams, handling simpler queries before handing off more complex issues to human agents. In addition, chatbots can also help consumers to find new products and services, and in some cases they have replaced Google Search as a tool for searching the internet.

Despite the enthusiasm for AI chatbots, the uptake of this technology has sometimes been problematic. Some users have come up with creative ways to mislead chatbots and get them to say controversial and problematic things, while there have been occasions where they might go astray on their own accord, recommending competitor’s products.

Obviously, brands need to avoid such problems if they’re going to become more reliant on AI chatbots, and this is where Palona AI thinks it can help. It’s partnering with OpenAI to use that company’s large language models as a foundation, but it adds its own model on top of that, which acts as a kind of “supervisor.” The idea is that it ensures chatbots will stay on topic, making them more aware of a brand’s identity and what they’re supposed to be doing.

With its technology, Palona AI says, it can help brands to develop customized AI sales agents capable of upselling products and providing personalized recommendations to consumers. It works with consumer brands to help them incorporate proprietary data such as their employee training manuals, frequently asked questions, product inventories and marketing materials, so they can fine-tune the models. In this way, its models are much better at representing brands and following their interests, the startup says.

Palona AI is led by co-founder and Chief Executive Maria Zhang, who previously worked as a vice president of engineering at Google, as well as a vice president and general manager of the AI for Products group at Meta.

Zhang told Bloomberg that one of Palona AI’s most compelling advantages is that its chatbots can encourage users to engage with them more, to the point where some will even ask it questions that they might be too embarrassed to say to a human agent.

She added that the company aims to “breathe life into brands” and deliver friendly and individualized experiences to every customer.

“Building relationships with customers requires a precise mix of human skills that even the best salespeople train for years to achieve,” Zhang said. “We captured that skillset and turned it into models and actions our AI Agents can consistently perform.”

Zhang’s co-founders include Chief Technology Officer Tim Howes, who is a former leader of Meta’s AI infrastructure team, and Chief Scientist Steve Liu, who previously held a similar role at Samsung.

Images: Palona AI