Issue 1 · July 14 · Issue 2 · July 21 · Issue 3 · July 28 · Issue 4 · August 5 · Issue 6 · August 19 · Issue 7 · August 26 · Issue 8 · September 1 · Issue 9 · September 8 · Current issue
A weekly read on what the AI noise is hiding.
Two-thirds of supply-chain digital investment now goes to AI. More than half of the officers writing those checks say they cannot see the return. The same week, a freight AI-agent company became a unicorn on the strength of 150 enterprise logistics customers. The spending is confident. The seeing is not.
The money flipped before the visibility did. Every stat sourced. The hype named as hype. One number worth acting on. Read the full brief below, free, no email required.
Read those in order. The money moved first. The returns are invisible to half the buyers. The capital markets are pricing the next wave anyway, and the adoption funnel says almost nobody is at the end of it yet.
Gartner published a survey of supply chain chiefs last Tuesday. Two numbers in it belong side by side.
67% of supply-chain digital investment is now allocated to AI. That is not an experiment line item. That is the budget.
55% of the chief supply chain officers spending it say they are unclear on the return.
Hold both at once. Two-thirds of the money, and more than half the buyers cannot see what it bought. Not “disappointed in the return.” Unclear on it. The instrument panel is missing, not just the altitude.
The usual reading of a number like that is “AI is not working.” That is not what it says. Some of those investments are almost certainly paying; some are almost certainly not; the 55% cannot tell which is which. Blindness and failure are different problems with different fixes, and the survey is measuring blindness.
Gartner’s own prescription gives the game away. It is not a model upgrade or a bigger platform. It is change management, sized to each initiative, explicitly linked to strategy. By 2030, Gartner predicts, organizations that do that will generate roughly twice the ROI of organizations running standardized legacy change methods. Twice the return, same technology. The difference is entirely in how the organization absorbs the tool.
That is an analyst firm concluding that the constraint is not the AI. It is the organization around it.
The same week the buyers reported blindness, the capital markets priced the next wave.
HappyRobot closed $150 million at a $1.2 billion valuation on August 4. The product is AI agents that absorb the phone calls, emails, and follow-ups that move freight. More than 150 enterprise customers, DHL and Kuehne + Nagel among them. This is not a lab bet; it is a bet on logistics back offices specifically.
Salesforce added its own numbers this week: agent deployments per business roughly tripled over 14 months, and seven of ten customer-service conversations now resolve without a human. Vendor numbers, counted on the vendor’s own platform, so apply the discount. But the direction matches what the spending data already said.
And the adoption funnel puts a floor under all of it. Blue Yonder’s survey of 678 supply chain leaders: 95% run machine learning or predictive systems, roughly a quarter use generative AI, 8% run agentic AI. Ninety-five to twenty-five to eight. Everyone has the old AI. Almost nobody has the new kind. The distance between those numbers is the market every vendor above is selling into.
The skeptic’s case got printed in the Financial Times this month, and it deserves the space.
ManMohan Sodhi of Bayes Business School argues that much of what ships under the “agent” label is rebadged rules engines, automation that would have been called workflow software five years ago. He also names the harder problem: individual productivity gains from AI tools consistently fail to show up as team output. The gain is real at the desk and missing at the department.
Sit with that second point, because it is not a complaint about the technology. Work compounds through handoffs. If the handoffs are broken, faster individuals just wait faster. A tool that speeds up every desk and speeds up no team is not underperforming; it is revealing where the actual constraint lives.
Which is the same place Gartner’s change-management prediction points, and the same place the 55% blindness lives. Three findings from three directions, one location.
Also worth saying plainly: nothing in this issue is a random sample. Gartner surveys its own client orbit. Salesforce counts its own platform. Blue Yonder surveys its own market. The convergence across them is the signal, not any single number.
Put the week together.
The money has already moved: 67% of digital spend. The visibility has not: 55% unclear. The vendors are shipping agents into the gap, and the capital markets are funding them at unicorn prices. And the one prescription the analyst community offers, tailored change management, twice the ROI, is a statement about organizational structure wearing a consulting term.
“Unclear on ROI” has a mechanical cause. Returns from AI show up as time: decisions reached faster, handoffs skipped, rework not done. If a company has never measured its coordination cost, the return lands in a ledger nobody keeps. The investment is visible because procurement recorded it. The return is invisible because nothing records where coordination time goes.
You cannot see the return on removing a cost you never measured. That is the whole 55%.
The fix is not another dashboard on the AI. It is a baseline on the organization: handoffs per decision, days between a call being made and anyone acting on it, meetings that exist so other meetings can happen. Companies that hold that baseline will be the ones inside Gartner’s 2x. Companies that do not will keep writing two-thirds of their digital budget into a fog.
Score yourself honestly before we do it for you: readiness.align-ify.com. 33 questions, free, ends with a real number.
One weekly read on what the AI noise is hiding. Every stat sourced, the hype named as hype, one number worth acting on. No spam, unsubscribe anytime.
Start by measuring the coordination tax you’re paying right now. Free, a few minutes.
Next Tuesday: another scan, same rule. Cut the hype. Show you what the noise is hiding.