really.ai

really.ai was born from a simple frustration. The original three co-founders were fantasy sports enthusiasts who wanted instant access to highlight clips—like an Alvin Kamara touchdown—the moment it happened so they could share it with friends. Rather than manually cutting clips from streams or waiting for official highlights, they started cutting games themselves and quickly realized the manual approach didn't scale. They discovered someone in academia who had written a book on machine vision and machine learning for video analysis, bought their work, and built the AI-powered system from there.

The technical challenge was immense. The team spent two years solving the hard problem of real-time video processing across different camera angles, elevations, and viewports. Balancing machine vision and machine learning with the speed required for live-game processing meant careful optimization—they had limited computational resources and needed clips ready in approximately three seconds. By the time Daniel Evans joined in 2019, the founders had built solid technology. The pre-pandemic business (2018-19) was doing around $17,000-18,000 per month, mostly serving collegiate athletic departments.

The initial customer base was collegiate sports programs—D2, D3, and some Division I schools—who recognized the value of automating their social media content distribution. George Mason University became an early marquee customer, purchasing rights to process 300 games per year across baseball, football, and basketball at roughly $40-50 per game (totaling around $12,000 annually). The pricing model was usage-based SaaS: customers paid per game based on their sport and season volume. By 2020, the company had found product-market fit and was positioned to scale.

COVID-19 devastated the business overnight. When spring and fall sports shut down in 2020, revenue dropped to nearly zero as customers had no games to produce content from. This could have been fatal, but Evans and the original investors understood this was temporary. They raised $1.8M in pre-seed funding at a $4M valuation in early 2020 from Stadia Ventures, a sports tech and esports fund, betting that sports would return. The bet paid off. By fall 2020, games resumed and customer renewals kicked in. What truly worked was the product's integration into athletic department workflows—once SIDs (Sports Information Directors) got used to automating their Twitter and social media distribution instead of manually calling trucks or hunting clips online, churn became nearly non-existent. Over two years with Evans, only two customers left. The expanded vision to monetize ad inventory and move beyond traditional sports into esports and high school NIL also opened new growth vectors.

From essentially zero revenue in mid-2020, the company hit approximately $110,000+ MRR by late 2021 with 128 customers spanning collegiate sports, professional teams, esports platforms, and high school programs. A major win came through a deal with Challenger Mode, a European esports platform with 2.5M users running 15,000 tournaments per month, which will feed content into ESPN-style highlight reels. Evans projected $3.1M ARR for 2022. The company is now raising $3.5M in what amounts to a seed Series A at a $12-13M valuation. Evans, who took no salary and structured his incentives entirely around company equity and exit, assembled an 11-person team (6 engineers) focused on expanding into esports and evolving the technology for new verticals. The founders retain approximately 40% ownership.