40% of enterprises will scrap AI agents - 3 ways to ensure yours don't fail
Summary
Gartner predicts that 40% of enterprises will decommission autonomous AI agents by 2027, primarily due to governance gaps identified post-production. Despite this, three digital leaders from Whoop, Fanatics, and Synopsys shared successful strategies at the recent Snowflake Summit for deploying AI agents. Matt Luizzi of Whoop emphasized establishing formalized evaluation frameworks and a robust semantic layer to scale agentic AI explorations, using Snowflake CoCo for analytics. Madeleine Want from Fanatics highlighted the critical role of high-quality, governed data and expert analysts to effectively coach agents, enabling increased accuracy and broader application. Sriram Sitaraman of Synopsys advocated monetizing data with AI agents for tasks like revenue reporting, noting improvements in result quality, time, and cost, while cautioning about the distinction between automation and autonomy.
Key takeaway
For Directors of AI/ML evaluating AI agent deployments, recognize that 40% of enterprise agents may fail by 2027 without proper governance. Prioritize establishing formalized evaluation frameworks and ensuring high-quality, governed data. Engage expert analysts to guide agent development and carefully differentiate between automation and true autonomy to avoid unforeseen costs and risks. Your focus on structured processes will drive measurable value and prevent costly decommissioning.
Key insights
Effective AI agent deployment requires robust governance, structured frameworks, and expert human oversight to ensure ROI.
Principles
- Formalized evaluation frameworks are crucial for scaling AI agents.
- High-quality, governed data improves LLM effectiveness.
- Distinguish between automation and autonomy in agent design.
Method
Implement repeatable evaluation frameworks, centralize data on platforms like Snowflake, and use expert analysts to coach agents in well-bounded contexts.
In practice
- Use coding agents like Snowflake CoCo for developer tasks.
- Deploy revenue agents for finance reporting.
- Embed agent APIs into third-party tools for broader access.
Topics
- AI Agents
- Enterprise AI
- AI Governance
- Data Monetization
- Evaluation Frameworks
- Snowflake CoCo
Best for: CTO, Executive, AI Architect, Director of AI/ML, VP of Engineering/Data, MLOps Engineer
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by News and Advice on the World's Latest Innovations | ZDNET.