Agentic AI Is a Team Design Problem, Not a Tool Problem
Agentic AI in enterprises refers to autonomous or semi-autonomous AI agents that perform end-to-end business tasks inside functions like finance, HR, legal, operations, and customer service, working across systems and data sources instead of staying confined to a single app or chatbot surface.
The most important shift in AI agents enterprise deployment is organizational, not technical: companies that embed AI talent inside business functions are pulling ahead. Uber’s decision to place 30 of its most AI-proficient engineers directly into finance, legal, and human resources teams to build domain-specific agents is not a side experiment; it is a blueprint for how back-office work will be redesigned. Meanwhile, retailers like Batteries Plus are proving that agentic workflows in finance HR and beyond only pay off when an enterprise has modern, cloud-native systems and a high-quality data foundation in place. In other words, agentic AI is exposing a harsh truth: you cannot bolt tomorrow’s automation onto yesterday’s architecture and expect real gains.
Uber’s Agentic Pods: AI Engineers as Temporary Colleagues
Uber’s “agentic pods” show what embedded AI teams look like in practice. The company placed 30 of its most AI-proficient engineers within finance, legal, and HR teams, then had them work shoulder-to-shoulder with staff for two weeks to observe workflows and build agents around real tasks, not theory. These pods have already run 16 times in two months, each time acting like a small, forward-deployed AI strike team.
This model directly attacks the messiest work: multi-step, cross-system processes that were previously stitched together by spreadsheets, emails, and tribal knowledge. Many of the duties engineers targeted, including financial pacing reports, required accessing multiple systems and heavy manual effort. Now, those reports can be produced in about 10 minutes instead of two days, and allocating capital across 150 cities has dropped from 15 hours to 30 minutes. “You can't automate them effectively by looking at process diagrams or documentation,” Praveen Neppalli Naga argued; you have to see how the work actually gets done.
Batteries Plus: Modernization as a Prerequisite for Agentic AI
Where Uber shows how to organize people, Batteries Plus shows how to organize systems. The retailer spent several years shifting its core platforms to a scalable, cloud-enabled, data-driven infrastructure, a move it describes as building an “AI-ready enterprise architecture” that positions it to scale AI across its business. This was not a vanity cloud project; it was a strategic bet that future automation would depend on clean, connected data.
That bet is already paying off. By upgrading ecommerce, point-of-sale, payments, product information management, enterprise search, and partner integrations, the company strengthened system reliability and built a foundational data layer that gives AI systems structured, contextual information to deliver accurate results. Among the gains: failed searches are down 25%, and automated self-service channels now contain 35%–40% of customer inquiries without human escalation. “The PIM transformation, data platform investments, and integration work created this knowledge foundation, which is now central to our Enterprise AI strategy,” Karthik Jambulingam explained. This is a quiet but sharp rebuke to enterprises hoping to deploy AI agents on top of fragmented, legacy stacks.
From Chatbots to Business Function Automation
Both Uber and Batteries Plus illustrate that the real value of AI agents enterprise deployment lies in deep business function automation, not surface-level chatbots. Uber’s pods go straight after finance and HR work, redesigning tasks like capital allocation and reporting into agentic workflows finance HR teams can trust. Batteries Plus, for its part, uses AI agents and assistants across store operations, commercial sales, and ecommerce, including voice agents tied into store phone systems, AI-powered product discovery, store call analytics, and a conversational assistant for associates.
These are not gimmicks. AI-powered voice agents now help with product discovery, service scheduling, and call routing while handing off to humans for complex cases. Intelligent product discovery tools help staff handle complex compatibility questions, and commercial AI agents autonomously identify high-potential leads and schedule sales appointments. This is business function automation that touches real customers, from buy-online-pickup-in-store journeys to commercial churn management. It shows that when enterprises commit to agentic workflows rather than isolated tools, they start to redesign how work itself is structured.
The Competitive Edge of Cross-Functional Agents
The pattern is clear: cross-functional AI agent deployment is becoming a competitive edge for enterprises willing to restructure both teams and architecture. Uber plans to keep using the agentic pod model and is forming a dedicated team to scale it, with the explicit goal of deeply understanding work, redesigning it from the ground up, and using AI to change how the business operates. Batteries Plus is already running AI across six major areas of its operations, from voice agents to agentic commerce, built on its knowledge-centric architecture.
The lesson for other enterprises is uncomfortable: you cannot expect transformational results from AI agents if you keep AI work centralized, keep processes static, and keep your systems stuck in the past. Agentic pods and embedded AI teams challenge traditional org charts; AI-ready enterprise architecture challenges legacy technology stacks. Companies that accept both challenges will turn AI from a lab experiment into a structural advantage. Those that do not will watch competitors quietly automate away the cycle times, error rates, and friction that still define their finance, HR, legal, and operations workflows.
