Category: Uncategorized
What happens when the agent is wrong in public
Happens: The plan for this should exist before the first deployment, not after the first incident.
Seasonality is not one pattern
Weather seasonality, calendar seasonality and promotional seasonality overlap in the same series, and a model that treats them as one gets each of them wrong.
The pilot that proves nothing
It ran on clean historical data, in a sandbox, against a metric agreed afterwards. It succeeded. It told you nothing about whether this works.
Agentic AI and the shift-handover problem
Three shifts, three sets of judgement, and a handover conversation that lasts four minutes. Most of what one shift learned never reaches the next.
Your ERP is not the system of record for decisions
It records what was done. The reasoning, the alternatives and the constraints that bound the choice are not in there, and there is nowhere else for them to go.
Reasoning that survives contact with a domain expert
The test of an explanation is not whether it sounds plausible. It is whether the person who knows the process can disagree with a specific part of it.
Inventory optimisation across a network is a different problem
Optimising each location independently produces a network that is simultaneously overstocked and out of stock.
What an agent needs to know before it can price anything
Elasticity is the easy part. The hard part is the list of prices you are not allowed to set, and why.
Scrap rates hide the decision that caused them
Scrap is recorded against a shift and a line. The decision that caused it was made three days earlier, by someone in another department.
Multi-agent systems make sense when the decisions genuinely conflict
Not because more agents are better. Because a purchasing agent and a production agent should argue, and a single model cannot argue with itself.