In 2025, AI’s impact was big, but this year is already shaping up to be bigger and bolder—and bringing even more upside for supply chain operators. We’ve already seen significant advances, with models becoming much more capable in early 2026.
After a year of energetic experimentation, enterprise leaders need to start scaling solutions that deliver proven value. The latest models are so capable that the limiting factor becomes the speed at which organizations can adopt them.
Boards are challenging CEOs to show that AI investments generate bottom-line returns. Successful companies are capturing compounding advantage by leveraging early breakthroughs—in efficiency, speed, and decision-making—to surge ahead of laggardly rivals.
This year, enterprise adoption will accelerate. ROI will take center stage as proofs of concept evolve into established best practices. It’s the year to get serious about AI—and build robust paths to value for both enterprises and customers.
5 key trends to watch
To succeed in this new era, enterprises must pay attention to five key trends:
1. LLMs will expand capabilities with better data
Large language models went from strength to strength in 2025, and we’ve already seen progress in 2026 driven by model post-training. The incredible progress of LLMs on popular tasks, such as coding, is largely driven by post-training model tuning.
At our company, we’ve learned that optimal results only come from models that are post-trained on specific supply chain data and use cases. Now, that insight should be applied across the supply chain, which is admittedly challenging given the private nature of operations.
Instead of using off-the-shelf tools, enterprises will increasingly adopt models fine-tuned on enterprise and supply chain data that natively understand business operations, because they are specifically trained with reinforcement learning.
2. Expert supply chain models will power the best agents
As these models are tuned with increasingly relevant and specific data, session histories, and tailored instructions—a process known as context engineering—they will power agents to solve more complex problems.
Augmenting models with real-time access to their knowledge graphs allows companies to build agents that are experts in their organization’s operations and supply chains and can intelligently optimize them.
Meanwhile, frontier firms will use world models (or models of physical cause-effect relationships) to teach AI to understand physical environments, yielding agents and robots capable of managing real-world tasks ranging from stacking shelves to picking orders in warehouses.
3. Agents will have increased autonomy to solve more complex problems
We’ve seen the growing use of agentic “vibe coding” tools such as Claude Code across the enterprise, and we’ll start seeing this in the supply chain in 2026. Custom-trained models and agents will also drive real-world gains. Organizations will leverage agents for knowledge work, helping agents and humans solve problems faster.
Expert agents will be given greater autonomy as they demonstrate their capabilities—and organizations will close the loop, allowing models to learn and make better decisions over time.
As they improve, agents will be deployed across larger swaths of the supply chain to solve increasingly complex problems. For example, if a pallet of goods arrives damaged, the agents won’t simply update the inventory to reflect the loss, but also reorder the product and coordinate delivery logistics. Agents will eventually autonomously monitor the supply chain and intervene when and where necessary, solving problems overnight.
4. Agentic shopping will bring challenges and opportunities
As agents scale, they’ll have a big impact. Agentic commerce is a $5 trillion opportunity to reshape the way people shop—and with the launch of OpenAI’s Agentic Commerce Protocol and Google and Shopify’s Universal Commerce Protocol, the sector is cleared for takeoff.
As consumers embrace agents, they’ll no longer default to giant online sellers, which will test supply chain leaders. Instead, consumers will instruct agents to find precise product configurations, low prices, or fast shipping, and channel sales through niche websites.
To support this, organizations will deploy their own agentic tools. At present, just 7% of supply chain leaders have real-time decision-making capabilities—but with fleets of AI agents managed via a central control tower, it’s possible to monitor volatile environments, anticipate demand surges, and respond quickly with optimized resources. Done right, that turns disruptions—including agentic commerce—into a driver of competitive advantage.
5. Agents will use a global context to operate across business units
The world will only grow more complex in 2026. Agents need to learn to think dynamically within the enterprise and across end-to-end functions. Instead of simply using AI to optimize warehouse operations, enterprises will use unified data and decisioning systems to provide a global context graph that connects their fulfillment, transportation, forecasting, and planning tools—driving value across their operations.
To accomplish this, agents must understand and work effectively across their broader ecosystems of customers and trading partners. The context graph will provide this global context underpinned by knowledge graphs and supply chain ontologies to accelerate, automate, and streamline business operations.
Get serious about ROI
Naturally, the evolution of AI brings challenges; we’re already seeing bad actors using AI for cyber-espionage, for instance. Economic risks, too, as the market continues to grow, but not every participant will thrive.
To succeed, supply chain organizations must develop disciplined strategies with real-world impact, and leverage both human capital and proprietary or industry-specific data and tools to drive compelling value.
Subject-matter experts and talent must know how to use AI tools effectively. They will also systematically assess AI solutions, determine which add value, and decisively scale up successful initiatives while terminating underperforming experiments.
Finally, organizations will need to think carefully about tech partnerships. Continuing to live in siloed systems will increasingly lock enterprises out of AI’s potential gains.
Organizations using unified data and decision-making on an agentic operation platform will increasingly outpace those continuing to operate across silos. The time to act is now. Leading organizations are charting a new course with AI, and finding partners who can grow rapidly alongside them in 2026 and beyond.
Chris Burchett is the SVP of Generative AI at Blue Yonder.