Playtest

Inspiration

ESPN loses billions of dollars (and steadily increasing losses) in revenue because advertisers know viewers watch and convert less and less on disruptive advertisements.

What it does

So we built a technology that can generate and integrate partner brand assets into video organically and in real-time.

How we built it

Roboflow, JavaScript, scraping and querying and getting responses from LLM to form brand identity.

Challenges we ran into

The accuracy and speed of our video recognition could be improved, and we're excited by the prospect of Omnimotion coming out which will be a major help. (It was co-authored by a Berkeley AI researcher!)

Accomplishments that we're proud of

Building video recognition models, coming up with a creative and USEFUL application of a new technology that we've found has direct value for a massive market.

What we learned

This industry serves players as large as the ESPN's and their partner Gatorades of the world. And is also extremely underserved by the latest advances in tech.

What's next for Playtest

We're going to solve creative ad opps for Twitch and YouTube video creators because they (a) have massive audiences but also (b) operate independently and make decisions far faster than larger live broadcasters. We've already attracted the attention of founders and investors from major eSports organizations and who count prominent folks in our space like Kevin Lin (Twitch co-founder) on their cap tables.

Built With

• gpt-3.5-turbo-16k
• javascript
• python
• scrapy