Agencies struggle to scale AI

More companies are using artificial intelligence at work, but many still cannot move it from small tests into everyday operations. Infor said April 22 that more than half of organizations in its new index struggle to scale AI beyond pilot projects. (infor.com) Infor said its Enterprise AI Adoption Impact Index surveyed 1,000 business decision-makers across the United States, United Kingdom, Germany, and France. The company tied the findings to new products in its Velocity Suite and a limited-availability Agentic Orchestrator aimed at governed deployment. (infor.com) At Disney, the gap between experimentation and scaled use looks uneven even inside one company. Business Insider reported April 23 that Disney is piloting an internal “AI Adoption Dashboard” for tech staff that tracks usage of tools including Claude and Cursor. (businessinsider.com) One Disney employee invoked Claude about 460,000 times over nine workdays in mid-April, according to the dashboard reviewed by Business Insider. Another employee told the outlet that numbers at that level likely reflect autonomous agents or automated workflows, not only manual prompting. (businessinsider.com) AI scaling is less about getting staff to open a chatbot than about wiring the tools into approved workflows, data systems, and controls. Infor said the biggest barriers are operational: companies are struggling with governance, picking the right use cases, and turning experiments into measured business results. (infor.com) That bottleneck shows up in broader U.S. data too. A Federal Reserve note published April 3 said Census Bureau business survey data showed about 18% of firms had adopted AI by the end of 2025, indicating that organization-wide deployment still trails the public attention around the technology. (federalreserve.gov) Inside companies, usage can spike faster than policy. Disney’s dashboard reportedly shows individual and team metrics such as requests and tokens consumed, giving managers a way to see who is using AI heavily and where costs may be concentrating. (businessinsider.com) Vendors are responding by selling more control layers around the models. Infor said its new tooling is designed to add industry-specific workflows, governance, and orchestration so companies can run AI in production systems instead of isolated demos. (infor.com) The result is a split-screen AI economy: some workers are generating hundreds of thousands of model calls in days, while many employers are still trying to decide where AI belongs, who approves it, and how to measure whether it is worth the cost. (businessinsider.com) (infor.com)