Building a strong data infrastructure for AI agent success
#AI agents #data architecture #business context #scaling AI #data governance #enterprise data #McKinsey report #agentic AI
📌 Key Takeaways
- Nearly two-thirds of companies were experimenting with AI agents in late 2025, but only 10% successfully scaled them.
- Scaling AI agents is hindered more by inadequate data architecture than by model shortcomings.
- Businesses need a modern data infrastructure that provides reliable data with business context to support AI agents.
- High-value data for AI agents is defined by business context, not by whether it is structured or unstructured.
📖 Full Retelling
🏷️ Themes
AI Adoption, Data Infrastructure
📚 Related People & Topics
AI agent
Systems that perform tasks without human intervention
In the context of generative artificial intelligence, AI agents (also referred to as compound AI systems or agentic AI) are a class of intelligent agents distinguished by their ability to operate autonomously in complex environments. Agentic AI tools prioritize decision-making over content creation ...
Entity Intersection Graph
Connections for AI agent:
Mentioned Entities
Deep Analysis
Why It Matters
This news matters because it highlights a critical bottleneck in enterprise AI adoption that affects nearly all industries. While companies are rapidly experimenting with AI agents, only 10% successfully scale them due to inadequate data infrastructure rather than model limitations. This affects business leaders, IT departments, and data teams who must prioritize data architecture investments to realize AI's promised value. The article warns that companies failing to build proper data foundations risk falling behind competitors in the AI race.
Context & Background
- McKinsey's 2025 AI report shows 88% of companies were using AI in at least one business function, up from 78% in 2024
- Nearly two-thirds of companies were experimenting with AI agents by late 2025 according to the same report
- Traditional data management has historically distinguished between structured data (seen as high-value) and unstructured data (seen as lower-value)
- Enterprise data architecture has evolved from basic databases to complex data lakes, warehouses, and now AI-ready infrastructures
What Happens Next
Companies will likely accelerate investments in modern data infrastructure throughout 2026-2027, with SAP and other enterprise software providers expanding AI-ready data solutions. Expect increased mergers and acquisitions as legacy companies acquire data architecture specialists. Industry conferences and certifications will emerge focusing specifically on AI data governance, while regulatory frameworks may develop around AI data quality standards.
Frequently Asked Questions
The primary failure point isn't the AI models themselves but inadequate data architecture. Companies lack systems that deliver business context reliably to both humans and AI agents, preventing effective scaling beyond pilot projects.
AI shifts value from data format (structured vs unstructured) to business context. High-value data for AI agents is defined by relevance, accuracy, and contextual richness rather than traditional structural characteristics.
Experts suggest the next few months to years are critical. SAP's Irfan Khan emphasizes that while AI's future is unpredictable, companies need immediate action on data foundations to achieve quick wins and remain competitive.
Only 10% of companies successfully scale their AI agents according to the article, despite 88% using AI in some business function and two-thirds experimenting with AI agents.
Enterprise leaders, CIOs, data architects, and business unit heads are most affected as they must prioritize and fund data infrastructure upgrades. The challenge also impacts AI implementation teams and end-users awaiting promised AI benefits.