ChatGPT's public launch left many companies scrambling to understand how the technology will impact work. Some companies took a wait-and-see approach: let early-adopter firms experiment, adopt early, and make costly mistakes. Others opted for pilot programs with certain business units or executives.
For many, banning ChatGPT from business seemed like a logical choice. The tech posed too many unknowns, like concerns about the data security of personal and proprietary information and, frankly, leaders were still worried that using the tool was akin to cheating.
Colgate-Palmolive took a different approach. Rather than restrict AI access, the company created a safe, internal environment for employees to explore using AI.
“The tools are broad based and extremely useful for everybody, for virtually everything that they do. Task number one is to make sure that the entire workforce feels comfortable with AI [and] understands that we want them to be using AI,” said Kli Pappas, global head of AI at Colgate-Palmolive. “Then all the other pieces can fall into place, and the world can shift around us.”
Early in 2023, the company built AI Hub, a central platform for its employees to learn about, use, and interact with AI.
“I own enterprise risk related to AI, and part of this was just a pure risk-based calculation,” Pappas said. “There's so much data [showing] that if you don't give employees access to tooling, then they will go outside of your network. When we looked at what was available for free publicly to people, we had to at least replicate what they could do outside.”
Additionally, Pappas’s team set up a tool inside the AI Hub where employees can build their own AI assistant, a custom GPT. Employees can give instructions and relevant data to the assistant, and then can use it at work.
The AI Hub enables employees to build any AI assistant they need. There are built-in governance processes for sharing. Assistants can be used by their creators, shared across teams, departments, or geographical regions through a governance framework. With more than 3,000 employee-developed assistants, some even rise up to be useful org-wide.
“This, at the time, was a revolutionary idea, because we were saying, ‘AI is not the domain of IT people or AI engineers and we’re in control of what gets built and [what] use cases are,” Pappas said. “There are thousands of…valuable things that people can do.”
Global reach. Colgate-Palmolive’s objective-setting assistant is homegrown. A member of the HR team built the tool that helps employees zero in on their goals. Trained on company values, priorities and strategies, as well as institutional knowledge about SMART goal-setting. The assistant coaches employees to develop better objectives.
Quick-to-read HR news & insights
From recruiting and retention to company culture and the latest in HR tech, HR Brew delivers up-to-date industry news and tips to help HR pros stay nimble in today’s fast-changing business environment.
The assistant began at the hands of one "very intrepid" employee, was tested by a group of roughly 20 colleagues, notified management, and now the tool is used across the globe.
“The person who built it is not in IT, is not a coder, is not a programmer, has no experience,” Pappas said. “All he did was just have the idea that this is a pain point and something that a chatbot can solve.”
Local focus. Not every pain point is shared by everyone; some are unique to a job function or location, and Colgate-Palmolive’s approach works for specific issues as well as broad ones.
A manufacturing plant in Greece, for instance, was outfitted with complex machinery and instruments from Germany, all equipped with manuals and documentation, in German. German is a fine language, but not exactly spoken much in Greece, and not by the employees who sometimes have to troubleshoot issues, and would be much more successful doing so in their native tongue.
The issue wasn’t big enough to get on the radar of IT or Pappas’s team. The plant manager didn’t need to file a support ticket or wait for an enterprise tool to roll out. Instead, he built a handful of assistants using the AI Hub and trained them on the German manuals and instructions. The assistants are available for his employees to use when an issue arises. He handled the problem locally. He never asked permission. He didn’t need to.
“It never made it out of that specific plant, because it was built for that plant,” Pappas said. “But this person literally had a dozen of these that he had set up just for his individual teams, and it was valuable for them.”
Data and impact. After a certain number of interactions with an assistant, Colgate-Palmolive requires employees to complete a survey about how the new assistant impacts their work. They cannot continue to use the assistant until they’ve complete it. Surveys assess benefits including time savings, quality of work, and creativity. Builders, managers and Pappa’s team collect qualitative and quantitative data about the impact of these AI assistants. Sometimes the data reveals such high usage that enterprise-level resources are poured into a particularly good idea.
“We’ve had literally thousands of these assistants get built in the past year and a half, somewhere in the range 3,000 to 5,000 that have been built across the company. It’s taken off in a way that we really didn't even imagine,” Pappas said.