The conversation about automation should include the person doing the task. This is the idea that allows you to read “The conversation that deserves more time: Automation and work” beyond the slogan. In automation and work, important decisions are rarely limited to a function: they organize relationships, distribute risks, and determine what a person can do when they need an explanation.
It is advisable to start without drama and without naivety. The conversation about automation must include whoever does the task can add value and, at the same time, open a new problem. The professional evaluation does not ask whether the technology is good or bad in general. It asks what purpose it pursues, under what conditions it works, and what consequences it leaves out of its main indicator.
The criterion appears when a promise becomes a verifiable situation. For the conversation about automation, it must include who does the task, that means describing users, context, information used and resulting decision. Naming these elements prevents words like innovation, truth, security or efficiency from replacing the analysis that each case needs.
Reconstruct the case before proposing the solution
Let's take a specific scene: a factory chose a robot without consulting operators who were aware of frequent traffic jams and maintenance risks. It is a test of how the system behaves outside of the demonstration, when rushing, incomplete information, crossed responsibilities, and people who have not received specialized training are involved.
The most useful diagnosis is this: late participation turns useful experience into resistance and makes correction more expensive. It allows you to investigate what rule, interface, incentive or organizational loophole made the outcome likely and why existing defenses did not detect it in time.
When talking about automation, it should include who does the task, separating fact, inference and decision avoids confusion. The fact can be verified; inference interprets partial signals; The decision applies a consequence. When all three layers are merged into one figure or automated message, it is difficult to provide context and almost impossible to know what needs to be corrected.
Also important is the temporal dimension of the conversation about automation; it must include whoever does the task. A measure provided during an emergency may be excessive months later. Correct information may become outdated. An immediate improvement can generate dependency. That is why any serious analysis must include duration, review, and an orderly way to end.
The exception does not invalidate the system: it reveals its border. In the case of the conversation about automation, it must include who does the task, it is interesting to know who is left out, what damage they suffer and what alternative they receive. If resolving an exception requires personal contacts or insistence for weeks, the formal route exists, but it does not work fairly.
Go from diagnosis to concrete practice
The priority action would be to incorporate the workforce from diagnosis to testing and recognize their knowledge. It can be applied first to a limited process, with a responsible person, defined resources and a date by which to decide whether to expand, correct or abandon it.
To evaluate the conversation about automation, it must include whoever does the task, I would take a reference prior to the change: total time, errors, complaints, abandonments and differences between groups. I would add brief interviews with those who use and attend to the process. Numbers detect patterns; The stories explain mechanisms that we still do not know how to explain.
The test must include a typical case, a borderline case, and a deliberate failure. This checks not only whether to incorporate the staff from diagnosis to the test and recognize their knowledge, but also what happens when data is missing, a provider fails, or a person disagrees. Rehearsing the fault calmly reduces the cost of discovering the procedure during an urgent situation.
Documenting the automation conversation should include who does the task and does not need to become an endless dossier. A page can collect purpose, scope, sources, responsible parties, risk signals, review channel and expiration. The essential thing is that someone outside the project can understand what was decided and why.
It is also advisable to preserve an alternative. It may be slower, but it must be accessible and capable of producing an equivalent result. For the conversation about automation, it must include whoever does the task, that output protects against errors, dependencies and cases that the design did not anticipate. It also provides an honest comparison on the true cost of the core solution.
Who is responsible when technology makes mistakes?
The conversation about automation must include whoever does the task requires identifiable responsibility. The supplier may operate a part, but the organization that decides to use it retains the duty to understand, monitor and repair. That responsibility needs material authority: access to records, budget, time, and the ability to stop a function.
An adequate explanation of the automation conversation should include the person doing the task answering five questions: what happened, what information influenced it, what consequence it produced, how long it will last, and how it can be reviewed. Offering only a category or a generic reference to the conditions does not allow you to understand or exercise rights effectively.
Early communication about the automation conversation should include whoever is doing the task is also a technical measure. He who knows the purpose may detect a strange result; Whoever understands the warning channel can communicate it before it escalates. Hiding uncertainty to protect reputation often increases damage and delays learning.
The conversation about automation must include the person doing the task: a digital decision is defensible when its purpose is clear, its limits are visible, and there is a person capable of listening, explaining, and correcting.
Let's go back to the opening scene. If this action had been applied—incorporating the workforce from diagnosis to testing and recognizing their knowledge—the result would have changed in a concrete way. That final question serves to avoid plans full of controls that do not address the real problem. It also forces us to recognize when a solution is disproportionate or arrives too late.
The analysis in “this case” offers no universal rule, but a discipline: describe before measuring, contrast before concluding, and repair before defending the tool. That practice produces more reliable systems and public conversations less dependent on promises, fears and automatic responses.
