The AI industry is rapidly evolving, with a plethora of new models emerging regularly, prompting companies to adopt these technologies to maintain their competitive edge. Tech consulting firm Searce reports that nearly 10% of businesses plan to invest an impressive $25 million this year in AI initiatives. However, amidst this financial commitment, concerns about return on investment (ROI) remain; according to Gartner, half of AI leaders struggle to measure or demonstrate the value of their AI projects.
Chetan Sharma, a former data scientist at Airbnb, argues that calculating AI ROI can be straightforward with the right tools. As the co-founder of Eppo, an innovative experimentation platform, Sharma enables clients to evaluate and tailor AI models for specific applications. Eppo not only offers model evaluation tools but also provides a comprehensive A/B testing platform for apps and websites.
“With the rapid launch of new AI models and significant investments from companies, A/B testing serves as a cost-effective solution to assess their performance without overspending,” Sharma stated. “Eppo empowers businesses to identify valuable models, allowing for smarter, more sustainable decision-making amid fast-paced innovation and rising costs.”
Eppo faces competition from several startups in the experimentation and A/B testing sector, such as Split, Statsig, and Optimizely, while industry giants like AWS, Microsoft Azure, and Google Cloud are enhancing their offerings in model fine-tuning and evaluation tools.
However, Sharma believes Eppo differentiates itself with features like its “contextual bandit” system, which automatically detects new variations of clients' websites, applications, or AI models, and intelligently evaluates their performance by directing increased traffic to them.
“Experimentation fuels growth by eliminating bureaucratic and often flawed committee-based decisions, tightly aligning initiatives with growth metrics, swiftly discarding ineffective ideas, and promoting successful concepts for reinvestment,” Sharma explained. “Eppo’s approach to live ‘online-eval’ testing of AI models effectively determines whether premium models enhance performance metrics.”
Since launching in stealth mode in 2022, Eppo has amassed “several hundred” enterprise clients, including notable names like Twitch, SurveyMonkey, DraftKings, Coinbase, Descript, and Perplexity. Alexis Weill, head of data at Perplexity, shared that Eppo has enabled them to “significantly scale” the number of concurrent experiments.
Investor interest is high, as evidenced by Eppo’s recent $28 million Series B funding round, led by Innovation Endeavors, with additional contributions from Icon Ventures, Amplify Partners, and Menlo Ventures. This new funding values Eppo at $138 million post-money and raises its total funding to $47.5 million. Sharma indicated that these resources will be allocated to enhance Eppo’s marketing efforts, AI experimentation capabilities, and analytics offerings, as well as to expand its market reach.
Based in San Francisco, Eppo currently employs 45 staff members and anticipates growing to 65 by the year’s end.
“The demand for efficient growth combined with the rise of AI has cultivated an adapt-or-die mentality, compelling companies to embrace experimentation,” Sharma noted. “As legacy vendors falter, many in the experimentation market have opted to develop large in-house teams rather than invest in external solutions. Given recent organizational shifts and layoffs, maintaining these in-house teams has become increasingly impractical, leading companies to turn to Eppo as a substitute for costly or abandoned internal projects.”
Ranking
