A new report reveals that enterprise AI adoption is still in its early stages, with only 25% of companies having deployed AI solutions in production. According to the State of AI Development Report by Vellum, based on interviews with over 1,250 AI developers, only a quarter of those who deployed AI have seen measurable impact. As reported by Taryn Plumb for VentureBeat, most enterprises are still in the strategy and proof-of-concept phase (53%) or beta testing (14%).
The study shows that document parsing and customer service chatbots are currently the most common AI applications. Companies that have seen success report competitive advantages (31.6%), cost savings (27.1%), and improved user adoption rates (12.6%) as primary benefits. OpenAI remains the leading model provider, though open-source alternatives are gaining traction. The report highlights significant challenges, including AI hallucinations, performance issues, and data security concerns. While AI development involves multiple departments beyond IT, with engineering (82.3%) and leadership (60.8%) leading the way, many organizations still lack technical expertise. Vellum CEO Akash Sharma emphasizes the importance of identifying specific use cases and proper evaluation processes for successful AI implementation.