Contenda was an accidental company. Originally, it was a hackathon project to help Twitch streamers retain subscribers. I had a hypothesis that we could run interventions on subscriber cohorts by sending people real life gifts (stickers) in the mail. The streamer we signed to run the project was Ludwig, who broke a Guinness World Record for most subscribers on Twitch in a given month.
Investors were banging on my door, demanding I quit my job at Facebook and build a company. By all means, it sounded like the start of another Silicon Valley dream. The reality was I had never been in management before and was woefully unprepared for all the jobs that come with being a founder.
Within the first 90 days of Contenda, I knew we couldn’t keep building the Twitch product. The business model that worked for Ludwig had tough margins, and Ludwig was a Guinness World Record holder. The tail was long and the margins only got worse. With plenty of runway, we looked to build a B2B product since we could cater to longer sales cycles.
Our advisor at the time, Cassidy, put us onto DevRel. DevRel had problems keeping up with content output and they had the budget to pay for it. GPT-3 was the most advanced language model in the market, but it was very flawed. It had lots of problems with hallucination, consistent quality, and there were no popular tools for building with GPT at the time. This seemed like the perfect problem for us to solve, given our runway and machine learning expertise.
We spent the next few months solving problems around quality and hallucination for GPT-3. I bold this because in startup land, this is an eternity. And worse, an eternity’s worth of work is wiped away in one announcement.
We originally tried to be positive. Our waitlist was growing at over 120% MoM because suddenly, everyone knew what AI was. Unfortunately, so was our competition. I’d guess that about 90% of the developments we’d made were superseded by the release of GPT-4. This meant that out of the box, competitors could launch a startup that competed with ours on little funding.
I hemmed and hawed for months, making a bad decision worse. I believed we still had the ability to pull off an enterprise sales motion. From sitting on multiple boards, I knew that industries were looking to implement AI across their companies. Surely, there was a place for us.
While companies were looking to implement AI, they were also looking for drastic cost savings. We saw layoffs in tech from tons of companies, and a lot of startups closed their doors for good. We brought in outside help on sales, which did improve our process, but it was still no dice. In the end, we had multiple deep conversations that pierced through all the relevant stakeholders who all seemed bought in, yet no deal was closed. It was agonizing to be so close and not get a single yes.
When I was finally ready to call it quits on our Contenda Studio product, I was heartbroken and exhausted. It felt like I ran a marathon, but didn’t finish the race. I took a week off, the longest break I’d taken from Contenda since we were founded.
The team had built a small side project called Brainstory on our API. They wanted something that would help them brainstorm ideas instead of create ideas, as generative AI promised to do. Someone added a feedback feature that allowed an author to share their Brainstory with someone else and it would guide them through giving feedback in the same open-ended way.
One day, a team member sent me a feedback request on a Brainstory about the company’s future. It was my first time using Brainstory, and it was completely magical. I didn’t understand it at all in description, but when I used it, it was… amazing.
With our Studio product, I was always trying to “wrangle” generative AI into producing what I needed. I was chasing after it, like a babysitter and a toddler. Generative AI tools always made me feel like the supporting cast.
With Brainstory, I was the main character. It made me feel sharp, witty. Normally, getting through tasks like responding to people’s Notion docs and emails could take me hours. I wanted to make sure I was thorough and read everything carefully. Brainstory helped me process and communicate ideas quickly.
There was also a sort of… afterglow of Brainstory. Not only did I feel productive and smart using it, but I noticed my attention span was deeper afterwards. It was easier for me to follow the many trains of thought in my head. Sometimes I feel like today’s apps wear down my attention span (looking at you 👀 TikTok). Today’s apps seem to always focus on delivering answers and satisfaction in seconds, under the threat of user attention waning. Brainstory doesn’t deliver answers, ever. You have to put in the work. In this way, Brainstory is like a personal trainer. Using it is like exercise, I notice that my mind feels clearer whenever I do it.
I’d like to think I’m a second-time founder now. This time, I’d like to make choices instead of accidentally ending up in them. Part of that is choosing to work on a product that I believe makes us better as thinkers in the world, not bystanders.
If any of this resonated with you, I want to hear from you. Come talk to us in our community Discord.