Estimates of the AI industryās market size range from $8T (McKinsey) to SpaceXās recent stratospheric estimate of $28T. No matter what size you believe the market to be, itās clear that even though 88% of large enterprises have already implemented AI in some form, almost none are close to realizing that full potential. The thing standing in the gap is trust.
Hereās the thing about that 88% of enterprises already using AI: most of them are using it to inform, not to act. That is a fine starting point, but itās not where the value lives. The real ROI in logistics comes when AI can execute automatically: resolving an exception, adjusting a forecast, booking a load, at a speed and scale that creates real competitive advantage. Getting there requires a level of trust most organizations have not earned yet. And trust is not a feeling. It is something you build deliberately, starting with the foundation the AI is built on.
AI you can trust starts with data you can trust. This matters more in supply chain than almost anywhere else, because the cost of failure is so high. A wrong answer is not a bad search result. It is a missed delivery, a stockout, a line down. So before you can trust what AI decides or does, you have to trust what it is built on. This is the first of three pieces on what trusted AI actually requires: trust in the data, then trust in the intelligence, then trust in the action. We start where everything else depends: the data.
The model is powerful. It is not enough.
Whether your teams use Claude, ChatGPT, or Gemini, they are using a large language model that reasons over public information. That is genuinely powerful, and it is also the wrong fuel for most business decisions. A foundation model has never seen your carriers, your lanes, your service commitments, or the dwell patterns at your busiest nodes.
Salesforce founder Marc Benioff put it well earlier this year: in the age of AI, the trusted data layered on top of a model is what drives real impact. The model is the engine. Your data is the fuel. And the specific fuel that makes AI useful is context.
Context is the situational knowledge that helps AI understand the relationships between pieces of information, so it produces the right output. It is the difference between a system that recites facts and one that gives you an answer you can act on.
Why AI without context creates noise, not outcomes
Most people have already felt this. You give an AI tool a job, you do not give it enough information, and it confidently goes off the rails. That experience shows up in the data. Research associated with Stanford and Microsoft Azure indicates that more than 80% of AI implementations fail because of insufficient data and context.
The inverse is just as striking. When AI is given the right context, it gets roughly 90% faster and about 70% more accurate. Think of it the way you would think about a new hire. Brilliant on paper, useless on day one without onboarding. Context is the onboarding. Without it, capability is just confident noise.
Supply chain context lives outside your four walls
Hereās what makes supply chain uniquely hard. The information that drives good decisions, including performance history, asset position, dwell patterns, and market shocks, mostly originates outside your organization. A pristine ERP cannot tell you how a carrier is behaving this week across everyone else’s freight, or where congestion is building three ports away.
That is why internal data alone, no matter how clean, cannot fuel trusted supply chain AI. You need the context that lives in the network, and you need it harmonized into something a machine can reason over.
The largest logistics data graph on earth
This is the foundation project44 has spent a decade building, and itās the richest source of supply chain context available in the industry. The scale is the point. The network maintains more than 5,000 unique APIs, captures over 1 billion customer-created events every day, and ingests 8 billion news, weather, and social events daily. Together that is more than 4 petabytes of data processed every month.
Itās powered by the largest carrier network in the world, with more than 267,000 carriers integrated across every mode and geography. That breadth is what turns raw data into context, because patterns only become meaningful when you can see enough of them.
Scale alone does not earn trust
The supply chain organizations making the most progress on trusted AI share a few qualities in how they built their data foundation. They get connected fast, because trust erodes when onboarding a new carrier takes months and context arrives too late to act on. They work with systems already in place, because a data foundation that requires replacing existing TMS, WMS, and ERP infrastructure will never reach the scale it needs to be useful. And they treat data completeness as a core capability, not a cleanup task.
A platform that cannot fill those gaps does not just have incomplete data. It has unreliable data, and unreliable data cannot support autonomous action. The organizations making the most progress have invested in closing availability gaps as a first-class problem, because AI is only as trustworthy as what it reasons over.
What trusted data looks like in practice
The outcomes are measurable when organizations have built a foundation they actually trust. Demand forecasting accuracy improves by as much as 25 percent, which translates directly to reductions in excess inventory and the carrying costs that come with it. Logistics costs fall 15 to 25 percent on average. And the response time to disruptions compresses from days to minutes.
The most visible shift is moving from reactive to predictive. An organization with trusted data does not wait for a customer to call about a missed delivery. The system flags the risk early, and action gets taken automatically, before the situation becomes a problem. That shift, from responding to disruptions to preventing them, is where the real ROI lives. It is also why Gartner projects that 60 percent of supply chain disruptions will be resolved automatically by 2031. That number is only possible if the underlying data is trusted enough to act on without a review cycle first.
This holds under pressure in ways that matter most. During the Iran conflict in 2026, shipping through the Strait of Hormuz was disrupted with little warning. The organizations that responded fastest were not the ones with the largest teams. They were the ones who did not have to question whether the data they were looking at was current and accurate. project44 offered port intelligence to customers for free so teams could reroute around congestion, and served as a source of truth on the situation for outlets including Bloomberg. That is what a trusted data foundation is for: it is most valuable exactly when the world is least predictable.
Key takeaways
- AI you can trust starts with data you can trust, not with a more powerful model.
- Context, the situational knowledge that explains how information relates, is the fuel that makes AI useful.
- More than 80% of AI implementations fail on insufficient data and context; the right context makes AI roughly 90% faster and 70% more accurate.
- In supply chain, most decision-driving context lives outside your organization, which is why network scale, fast connection, interoperability, and data quality matter.
Trusted data is the foundation. But data alone does not make decisions. In part two, we turn to what sits on top of it: trusted intelligence.
See how the largest logistics data graph fuels trusted AI in Movement. Talk to an expert.
