As the AI industry rapidly evolves, businesses are racing to adopt new models to maintain a competitive edge. According to tech consulting firm Searce, nearly 10% of companies plan to invest up to $25 million in AI initiatives this year. However, despite significant spending, many leaders struggle to measure the return on investment (ROI) from their AI projects. Gartner reports that half of all AI leaders are uncertain about how to calculate or demonstrate the value of these initiatives.
Chetan Sharma, a former Airbnb data scientist and co-founder of Eppo, argues that determining AI ROI can be straightforward with the right tools. Eppo is an experimentation platform that allows customers to evaluate and customize AI models for specific applications. In addition to its AI model evaluation suite, Eppo offers a comprehensive A/B testing platform for apps and websites, providing businesses with a cost-effective way to assess AI effectiveness without overspending.
"As new AI models emerge weekly and companies pour millions into them, A/B testing offers a way to evaluate their effectiveness economically," Sharma told TechCrunch. "Eppo helps companies identify which models truly deliver value, enabling smarter, more sustainable decisions in a rapidly innovating and costly environment."
Eppo competes with several other experimentation platforms, including Split, Statsig, and Optimizely, as well as major cloud providers like AWS, Microsoft Azure, and Google Cloud, which offer model fine-tuning and evaluation tools. However, Eppo distinguishes itself with features like its "contextual bandit" system. This system automatically identifies new variants of customers' websites, apps, or AI models and dynamically explores their performance by gradually increasing traffic or load.
Sharma emphasizes that experimentation accelerates growth by eliminating bureaucratic decision-making, tightly linking initiatives to growth metrics, and quickly discarding bad ideas while promoting successful ones. Eppo’s approach to live “online-eval” tests of AI models provides clear answers on whether premium models improve key metrics.
Since its stealth launch in 2022, Eppo has gained "several hundred" enterprise customers, including prominent names like Twitch, SurveyMonkey, DraftKings, Coinbase, Descript, and Perplexity. Alexis Weill, Perplexity's head of data, credits Eppo with enabling the company to significantly scale the number of experiments it conducts simultaneously.
Investors are taking notice. Eppo recently closed a $28 million Series B funding round led by Innovation Endeavors, with participation from Icon Ventures, Amplify Partners, and Menlo Ventures. This latest round values Eppo at $138 million post-money and brings its total funding to $47.5 million. Sharma plans to use the funds to enhance Eppo’s AI experimentation capabilities, expand its analytics offerings, and scale its go-to-market efforts.
With 45 employees based in San Francisco, Eppo expects to grow to 65 by the end of the year. Sharma highlights that the pressure for efficient growth, coupled with the rise of AI, has created an "adapt-or-die" mentality that drives companies to embrace experimentation. He notes that due to the limitations of legacy vendors, many companies previously relied on large in-house teams to build custom experimentation solutions. However, with recent layoffs and increased employee movement, these in-house teams are no longer sustainable, leading companies to turn to Eppo to replace costly or abandoned in-house systems.
