Pairing mentors with their dream startups through an AI-driven applicant tracking system
Helping Endeavor’s team of mentors find their perfect match among the massive pool of startups using an AI that helps them find The One.
what we did
The Engine behind startups
Endeavor is the leading global community for high-impact entrepreneurs. Their focus is inspiring, mentoring, and investing in the best startups to help them scale smartly.
We’ve always been fans of working with startups, and since Endeavor has been an active partner for our clients, this was a great opportunity to help make the startup ecosystem a little bit better.
Applying is tough
Endeavor handles a ton of applications from entrepreneurs each year, which are handled by an expert team of coaches and mentors, powered by a heavily customized Salesforce CRM.
But, as customized as it was, the CRM wasn’t able to keep up with the team’s growing needs. Problems arose. Missing files, out of date information, data spread across a dozen Google Docs documents — the amount of documentation on every startup was getting unwieldy, all of which was making the mentors’ job harder and more time consuming.
Building the platform
We set out to help the leadership team at Endeavor spend more time mentoring entrepreneurs and less time navigating applications.
Since Salesforce comprised the core infrastructure across multiple teams, we needed to carefully migrate to a new, faster system without disrupting the workflow of the mentors.
Making the onboarding process simple
In order for a mentor to properly analyze a company — they need to sift through a ton of information. In order to get all that documentation, Entrepreneurs are asked to supply anything and everything that can explain how and why a startup exists.
This includes information spanning from the Founders’ background to the financial forecasts of the company — and the process of finding and sending this information could easily take a few days to procure, organize, and send. Asking this from a Founder working 100 hours a week is a tall order.
In order to make this task more achievable, we redesigned the entire onboarding process so it could be done over multiple days. And we greatly reduced the amount of information that needed to be provided at every step. By turning this into a step-by-step process instead of a big one-and-done, applicants found it much more convenient to supply this necessary information.
This helped the onboarding process move faster, and take less time to complete.
Finding the one
Pairing the right mentor with the right startup is key to a successful and growing partnership. But it is an art form, and with thousands of startups in the community, it can be difficult to match people accordingly.
We tackled this from
an unexpected feature:
The search engine.
With some help from OpenAI and its wonderful Embeddings API we created a really smart search engine that could narrow down applicants and find exactly the teams they were looking for.
Now, instead of sifting through hundreds of profiles to get a single match, startups could type in, “I want a mentor for a fintech startup in the Bay Area that has a large background in enterprise integrations,” and get high quality matches in a couple seconds.This was a gamechanger for the team. And what started as an experiment by the product team, quickly became useful for the entire staff — from finding the fastest growing companies to discovering the perfect startups for an investors’ portfolio.
Changing the way people leave feedback
Asking for feedback is generally a futile exercise. When faced with a long form and broad questions, people usually close the page and move onto something more pressing.
We needed to get a ton of feedback from expert mentors, so we redesigned the entire feedback and selection system around the mentors’ schedules.
We simplified the questions so the bare minimum responses would suffice. This turned the feedback system from a set of unending surveys into lots of simple questions.
By carefully designing the questions and collecting more accurate data around them, we were still able to get high quality answers a lot faster than before.
A great use case for AI
By leveraging OpenAI's GPT4 we were able to do complex searches in a very noisy dataset, allowing people to find everything they were looking for.
The application process became faster to complete, and easily accessible — with all information living in one place.
A more reliable product
We increased reliability by implementing continuous integration and testing, and allowing the product team to test new features quickly to see what the Mentors responded to most.
"Working with Aerolab was such a refreshing experience. Compared to a typical agency-client relationship, they created an atmosphere of unity and mutual purpose. It felt as if we were one team united towards a common goal."