How Endeavor is using AI to build a matchmaker for entrepreneurs

Back in the summer of 2023 when everyone was talking about the incredible potential of ChatGPT and other generative AI tools, we at Endeavor noticed a pattern in our staff Slack channels: fully 60% of messages were team members requesting help finding a mentor to match with an Endeavor Entrepreneur looking for support.

This isn’t unexpected. At its core, Endeavor is a network. We connect entrepreneurs around the world to mentors and capital. Our network contains over 2,600 entrepreneurs and 5,000 mentors, plus additional board members, investors, and former employees. That’s a total of more than 10,000 people curated over 25 years.

The sheer size of our network means we are never more than two degrees of separation away from the perfect mentor to help any given entrepreneur. It also means our staff have needed huge amounts of time and assistance to sort through all the data we have collected to find that person.

Before AI our account managers relied on each other, their own memories, and traditional searches of our CRM. But with conversations raging about advanced technologies, we wondered what would happen if they could lean on AI as well? The idea for Endeavor Brain was born.

Endeavor’s bedrock principles for AI

Endeavor Brain is our name for our holistic advanced-technology strategy. It’s still in the early, experimental stages but it’s already yielded key insights about how high-touch organizations can use AI to better serve clients and customers, as well as what pitfalls and roadblocks they may run into along the way.

Looking at our Slack channels, we realized an initial way to push towards these goals was to use Gen AI to make institutional knowledge faster and easier to access for our staff, improving our skill as matchmakers.

The project was built on our existing, multi-year collaboration with digital product agency Aerolab. As we started out developing our first AI product with them we kept a few fundamental principles in mind.

First, as Endeavor Global CFO Margarita Chavez puts it, “Our main product is a curated experience for entrepreneurs. That is what we provide. So AI and any platform that we do is to enhance our capability to do so.”

We want to build AI tools that lead to more and deeper human connections. Rather than dream of a future where technology cuts out the middleman, allowing entrepreneurs to somehow interact with a platform or tech product, we want AI to take time-consuming busy work off the plate of our team members, freeing up energy and hours to devote to core, high-value tasks like building relationships, sharing knowledge, and expanding their own skills.

Building AI for great human connection

Like any service or consulting organization, we worry about maintaining our knowledge base when experienced team members move on to become entrepreneurs themselves or pursue other opportunities. And like many other knowledge-based organizations, we try to counteract this natural loss by recording information on our network, relationships, and learnings in our CRM. Traditional methods of searching through this data using tags and keywords was only so helpful.

That’s why our account managers were spending so much time asking colleagues for mentor ideas on Slack. It was also why a handful of top-of-mind names kept coming up again and again, while other equally knowledgeable and skilled mentors were sometimes overlooked. Better options were often available but were hard to surface with traditional tools.

As Aerolab co-founder and CTO Roberto Gonzalez explains, “Endeavor, because it has a very high-touch, one-to-one relationship with all the entrepreneurs and mentors, most of the data isn’t really structured. It’s available as meeting notes, recommendations, internal messages, all stuff that machines can’t understand really well.”

Our first Endeavor Brain project has been developing a chatbot that can pull from all these sources to provide staff with fast, useful mentor suggestions, saving them time, reducing any bias caused by ease of recall, and widening the pool of potential mentors they can consider.

Teething problems and learnings

With any emerging technology like Gen AI, humility is key. As we have gone through the process of internal engineering tests, then a limited alpha test, and now a wider organization-wide beta test of our first Endeavor Brain experiment, we have seen not just the potential of this technology, but also the pitfalls and concerns that any organization developing a similar product needs to consider.

Google’s chatbot infamously recommended putting glue on pizza, and Endeavor Brain, too, has sometimes produced AI “hallucinations,” suggesting superbly qualified mentors that just don’t exist.

That includes support in the new skill of how to talk to AI to get the best answer you want, a.k.a. prompt engineering. It also means stressing to users that Endeavor Brain is only a research tool. Human investigation and judgment are essential.

The Endeavor Brain project has also underlined the importance of maintaining high-quality records on our network and all our interactions with it. We are now working to train teams across 40+ countries to capture as much data as possible, in the correct formats, to ensure this and future AI projects provide as much value as possible.

When I think about Endeavor’s entry into the world of AI, I often think of a tweet by author Joanna Maciejewska that recently went viral. “I want AI to do my laundry and dishes so that I can do art and writing, not for AI to do my art and writing so that I can do my laundry and dishes,” she wrote.

At Endeavor, the equivalent of “doing the laundry and dishes” is trawling through the CRM and crunching data. Our “art and writing” is connecting entrepreneurs in underserved markets to the support they need to make an even greater impact on the world.