State of AI Agents: The Numbers Are In, and They're Bigger Than Expected
Why it matters: LangChain surveyed over 1,300 professionals and found that 57% already have AI agents running in production environments—not experiments, not proofs of concept, but actual deployed systems. Customer service leads the pack at 26.5% of use cases. But here's the sobering detail: 32% cite quality as the biggest barrier to scaling further. This tells me we're in a fascinating middle ground. Agents are clearly delivering enough value that organizations are pushing them live, but we haven't cracked the consistency problem yet. If you're holding off on agent deployment until things are "more mature," you may be falling behind competitors who are learning by doing.
Read full articleAI Spending Is Recession-Proof: Leaders Commit $124M Despite Economic Uncertainty
Why it matters: KPMG's Q4 AI Pulse Survey reveals something remarkable—67% of business leaders say they'll maintain AI spending even if a recession hits. The average projected deployment? $124 million over the coming year. But the real insight is buried in the details: leaders are "professionalizing" their agent systems with proper governance and observability frameworks for multi-agent deployments. This signals a maturation of the market. Early adopters learned the hard way that throwing agents at problems without oversight creates chaos. The organizations winning now are the ones treating agent infrastructure like any other critical business system—with monitoring, accountability, and clear operational boundaries.
Read full articleMicrosoft Goes All-In on Retail Agents: From Merchandising to Fulfillment
Why it matters: Microsoft isn't just adding AI features—they're building an entire agentic layer for retail. The announcement covers merchandising, marketing, store operations, and fulfillment with purpose-built agents for each. The standout offerings include Brand Agents for Shopify merchants and catalog enrichment agent templates in Copilot Studio. What catches my attention is the breadth of the approach. Microsoft is betting that retailers want unified agent ecosystems, not point solutions. For mid-market retailers especially, this could dramatically lower the barrier to sophisticated automation. The question becomes: do you build custom agents or adopt Microsoft's opinionated framework? Both paths have tradeoffs worth considering carefully.
Read full articleSAP's Order Reliability Agent: Fixing Problems Before Customers Notice
Why it matters: At NRF 2026, SAP unveiled what they're calling an "agentic operating system" for retail, with two flagship offerings: Retail Intelligence and the Order Reliability Agent. The latter proactively identifies potential order issues and resolves them before they impact customers. Think about that for a moment—an agent that monitors your entire order pipeline, spots a delayed shipment from a supplier, and automatically reroutes inventory from another distribution center. No human intervention required. This is the promise of agentic AI finally becoming real in enterprise software. SAP embedding this directly into their retail core means these capabilities will become table stakes, not competitive advantages, within 18-24 months.
Read full articleGoogle's Universal Commerce Protocol: A New Standard for Agent-Based Shopping
Why it matters: Google introduced the Universal Commerce Protocol (UCP), designed to work alongside existing standards like MCP, A2A, and AP2. The goal? Enable AI agents to complete actual shopping transactions—not just research products, but purchase them. Lowe's, Michaels, and Reebok are already deploying branded AI Business Agents within Google Search. This is Google making a land grab for agent-mediated commerce. If UCP becomes the standard way agents interact with merchants, Google controls a crucial piece of infrastructure in the AI economy. For businesses, this raises strategic questions: should you build agents that work with UCP? What does SEO look like when agents, not humans, are doing the shopping? These aren't hypothetical questions anymore.
Read full articleMcKinsey Warning: Don't Sacrifice Your ERP for AI Shiny Objects
Why it matters: This piece highlights a McKinsey warning that organizations are investing in AI at the expense of core ERP capabilities—and that's a problem. The good news? Vendors like Infor and Oracle are showing how AI agents can enhance rather than replace ERP systems. The 2026 architecture emerging here transforms manufacturing ERP into "systems of autonomous operations" where agents handle multi-step workflows while ERP provides the foundational data and processes. My take: this is exactly right. I've seen too many organizations treat AI and their existing systems as an either/or choice. The winners will be those who figure out how agents and ERP work together, with each doing what it does best.
Read full articleNetworks That Think: Agentic AI Transforms Infrastructure Management
Why it matters: NTT DATA explores a less flashy but critically important application of agentic AI: enterprise network management. AI agents now handle log intelligence, process monitoring, and even carrier coordination. The shift is from reactive (fixing problems after they happen) to autonomous (anticipating and preventing issues before users notice). For IT leaders, this represents a fundamental change in how infrastructure teams operate. Instead of staffing for incident response, you're staffing for agent oversight. Instead of alert fatigue from monitoring systems, you're reviewing agent decision logs. It's a different skill set and a different organizational model. Start thinking about this transition now.
Read full articleLooking Ahead
The theme of this week is unmistakable: agentic AI is no longer emerging—it's arrived. With over half of surveyed organizations running agents in production and enterprise giants embedding agent capabilities into core platforms, the early-mover window is closing. What I'm watching for in the coming weeks: how quickly Google's UCP gains merchant adoption, whether Microsoft and SAP's retail agent suites start competing for the same deals, and most importantly, how organizations solve the quality and governance challenges that remain the primary barriers to scale. If you're still in the planning phase with AI agents, the data suggests it's time to accelerate. Your competitors certainly are.