David is The Brunch Guy, both on Twitter and in real life. Pre-pandemic, David built a community of founders in NYC and would host dozens of brunches. Past themes ranged from uncomfortable questions to your relationship with death. Across 50+ brunches there was one consistent rule; everyone who comes to the brunches can’t know at least half the other people.

The most recent "founders brunch"

The most recent "founders brunch"

MeetSunday launched on June 8th, 2020 to 45 beta testers.

MeetSunday launched on June 8th, 2020 to 45 beta testers.

When the world shut down in April, 2020, David approached Rob with the idea to deliver the same level of connection that people would experience at his brunches, but over a digital medium. Rob, who has been an avid gamer since he was 5 years old, was drawn to the concept. We wanted to preserve the magic of meeting someone new and help our friends through the wide-spread loneliness, so we launched MeetSunday a few weeks later.

The premise was simple, members in the community would receive a text on Thursday and would sign up for a time to meet someone else in the community. David and Rob would match members who signed up and on Sunday a script would create group chats between those members.

At first, we randomly matched people and had them meet on Sundays over Zoom.

At first, we randomly matched people and had them meet on Sundays over Zoom.

Despite being a fun way to meet people, we learned that these conversations lacked purpose. Members didn't feel like there was a good reason to meet the other person. We experimented with matching people based on compatible personality traits and shared interests. We also included icebreakers so they had topics to discuss on the call.

https://s3-us-west-2.amazonaws.com/secure.notion-static.com/de4dbbc5-0a91-4d49-a52b-df794ec7f1d5/Screen_Shot_2021-07-29_at_7.04.00_PM.png

Members finally had a profile to customize their preferences - the introductions got better and better.

We fully automated the introductions by the end of August 2020 and found that about 20% of our community members opted in on a weekly basis. MeetSunday had become the primary social life for this subset of users.

https://s3-us-west-2.amazonaws.com/secure.notion-static.com/46482df9-112d-4ce0-b2f0-8124990a318c/Untitled.png

https://s3-us-west-2.amazonaws.com/secure.notion-static.com/c7f7eaf8-ae93-4656-b556-ae87494bfb14/IMG_3146.png

As the community grew, we ran into an issue. As we went from 10s to 100s of users, it became harder to maintain the quality of matches. We began by interviewing members and asking them “why did you join MeetSunday?”. Almost universally, their answers involved the concept of trust. People who we knew directly joined because they “trusted our network”. People who were second-degree connections to David and Rob mentioned “trusting the person” who recommended they join MeetSunday. If we couldn't vet everyone joining, then the product would lose value.

https://s3-us-west-2.amazonaws.com/secure.notion-static.com/d4abe0ae-57ab-4d1a-a323-226f99af5fa7/Screen_Shot_2021-07-29_at_6.35.38_PM.png

It is at this point, that we decided that growing and monetizing our own community was not the right strategy. After 50+ interviews with community managers, David and Rob decided to pivot. We took the introduction software we built for MeetSunday and let other communities use it for their own members.

We shutdown MeetSunday in November 2020, officially incorporated as Intros AI, and got to work.

Landing page in December 2020

Landing page in December 2020

By January 2021, we landed our first clients (August, Visualize Value and The University of Michigan CogSci department).

After the first month, we discovered an issue. Each community had their unique own preferences for how members should match. Some communities wanted to match members based on professions, years of experience, and skill sets. Others wanted to match members based on the type of activism they spend their time on.

Rather than building a one-size fits all tool, we needed to let communities customize their own algorithm for member matching. So, we set out to build the most customizable algorithm on the market. Not only did we let community managers choose the factors, we let them choose the weights for questions too. Community managers could say "I want 20% of the matches to be made based on age and 10% based on their favorite type of book and 50% based on their favorite color."

Community manager customizing questions for their members

Community manager customizing questions for their members

https://s3-us-west-2.amazonaws.com/secure.notion-static.com/5cd4505b-4997-4dd9-b7ef-0d5099912b28/Screen_Shot_2021-07-29_at_7.27.14_PM.png

By March 2021, the vision for Intros started to take shape. Quite a lot’s happened since then, be sure to reach out if you’d like to hear more!