Agentic AI Isn’t a Tech Challenge — It’s an Organizational One

Are people your “most important asset?”   

The phrase is a trope, and as GDP grows while hiring stalls, it sounds increasingly hollow. Since ChatGPT’s November 2022 launch, the S&P 500 is up 73% while total US non-farm job openings are down 32%. Casual observers could blame AI for taking jobs, and even if it’s not as simple as that, the fact remains that AI agents are increasingly able to handle work that is done by people. 

Layoffs are in the news as companies including Amazon, Meta, UPS, and Target trim hiring plans and headcount while investing in automation and AI. Productivity expectations are baked into current stock prices. Plus, consumers trained by disruptors like Shein and BYD expect things to be fast, cheap, and good. 

Don’t Optimize Silos, Build Power Threads to Win End-to-End

The question COOs are asking is how to organize their teams, including both humans and agents, to make the most of a technology breakthrough able to completely rewire business operating models. Why, for instance, keep an S&OP process that takes six weeks and three four-hour meetings of 20 people when an agentic system overseen by a handful of planners gets better results in hours? Always-on planning like this is less a business process than a decision support capability for everyone. 

More important, how should work be organized to ensure that upstream decisions are made with an understanding of downstream supply chain impacts, particularly between product design, engineering, and supply chain operations? The reverse is also true. How should work be designed backwards from the customer to ensure that sales promises are kept without massive inventories or expensive logistics? 

The answer is something we call Power Threads. These are workflows where multiple AI use cases connect functional areas to create multiplier effects that transform end-to-end operations. We have analyzed over 100 such Power Threads and identified the top four: 

• Inventory & Replenishment Optimization – connecting consumer transactions to demand, supply, production, and materials planning, and back to last mile, inventory deployment, and first mile.
• Barrier-Free Demand to Product – connecting marketing insights and product availability to product development, supplier management, demand planning, and finally, inventory deployment.
• Manufacturing Excellence & Automation – connecting customer sentiment analysis and product diagnostics to product development, and back to category intelligence, production quality, and manufacturing efficiency.
• Source to Contract & Plan – connecting product availability to IBP and materials planning, and back to e-sourcing, category intelligence, materials sourcing, and contract lifecycle management.

New operating models designed to exploit these cross-functional workflows are in flight now. Unilever hit 98% on-shelf availability with three-day lead time and 12% growth with a redesigned customer planning and fulfillment system using agents. Walmart reduced time-to-market for new fashions by 75% with its agent-supported trend-to-product system. General Motors saw ten-fold increases in defect detection with an AI-assisted quality control system. In these cases, AI agents are part of the team, not just tools. 

Ambitious transformations built on agentic systems are leading to the creation of new roles, new performance metrics, and radical improvements in productivity. But they don’t work at all without the right people leading the way. 

Size Matters

IBM recently published a study on AI which found that 66% of respondents saw “significant” productivity gains in operations. A quarter of those credit AI with “fundamentally changing their business models.” Even more interesting, 74% of big companies said they’d had success vs only 55% of SMBs. 

The data confirms a truth about agentic AI’s transformative potential: it works best in legacy companies with organizational complexity and inertia. This is especially true for big CPG, retail, and automotive companies with oligopolistic industries and long-tenured corporate staff. Coordination and reconciliation processes between functions, business units, and geographies work like a ratchet, ticking up more easily than down. Automating this work, as agents can, is a massive unlock. 

Human-Machine Teams Are the Future

And yet, success stories always have one thing in common: great people. This includes Translators who bridge ops and tech, superusers who train and tune agents in roll-out, and leaders who solve problems, track performance, and keep the faith. 

These people are your most important assets.