Opinion
When commercial promises exceed technical capability
A technological promise may rest on something that already works or announce a capability that still needs to be built. Both ask for trust, but on different grounds. That difference matters because a presentation can bring together, in a single experience, what is available, what works under certain conditions and what is only beginning to be imagined. To the person watching, the boundaries are not always visible.
What interests me is the moment when those differences disappear from the conversation. A possibility begins to be described with the certainty of a result; the conditions that produced it fall out of the explanation. The promise then acquires an authority the work cannot yet sustain. The problem is not excessive imagination, but a failure to explain how much remains to be done.
What appears and what holds
Consider a conceptual example: a system receives documents and produces a report. During the demonstration, the files are complete, share a format and contain the necessary information. The report is clear and the sequence looks straightforward. Everything shown may be real. The question is what that experience lets us conclude: we have seen the system perform a task under certain conditions; we do not yet know what happens when those conditions change.
Now imagine that a page is missing, two documents contradict each other or a figure appears without a reporting period. The task is no longer simply to produce text. The system needs to recognise what it can resolve and what it should return to a person. A demonstration built around complete documents does not answer that second question. Nor did it have to, provided its scope was made clear.
An interface can help us understand a possibility without establishing every capability we associate with it. That is its strength, and also the point that deserves attention. If a report is well presented, it is easy to assume equal care in source selection, error handling and review of its conclusions. Yet each of those operations needs an explanation of its own.
The future spoken in the present tense
In that example, saying the system “produces reports from these documents” describes what was observed. Saying it “handles report production” may imply a much broader scope: receiving varied material, recognising gaps, preserving traceability and allowing corrections. The change looks small within a sentence. For someone reorganising their work around it, the difference may be decisive.
This creates a shared responsibility, though not a symmetrical one. The builder knows decisions and dependencies that the observer cannot infer and needs to make them understandable, even when a more precise explanation narrows the promise. In our example, the builder should say whether the report comes directly from the system or whether someone selects documents, resolves contradictions and corrects the delivery. The person evaluating can ask about those tasks, but their silence does not turn hidden work into an automated capability. The responsibility is to provide an account on which another person can organise their work without later having to discover which parts of the solution depended on someone.
This is why I find the opposition between believers in a technology and its sceptics insufficient. In the example, someone could recognise the report’s value while considering it too early to delegate the whole process. That position allows progress: use what works, examine what fails and decide what should be built next. Trust can have a precise scope and still be trust.
The value of a limit
A clearly explained limit provides information for making a decision. Knowing that the system requires complete documents allows the material to be checked before use; knowing that a contradiction requires human judgement allows that responsibility to be assigned. The restriction becomes a working condition instead of an abstract reservation. It also gives improvement a specific meaning. If the system later recognises missing pages, we can discuss which new task it performs and which review remains necessary.
The limitation therefore needs to accompany the claim it qualifies. If a figure without a reporting period needs confirmation, the delivery should make that question visible before the report is used for a decision. A generic warning at the end leaves the reader to discover which conclusion it affects. Explaining where a person intervenes, what they correct and what happens if that review is absent defines the scope of automation. When those conditions change, the account of what the system offers needs to change too.
Imagination remains indispensable. The initial report may open a path that nobody yet knows how to follow to its end. A responsible promise makes it possible to recognise what already exists, the conditions under which it works and the work proposed for later. Each part asks for a different kind of trust. Its value is that participants can accept that path with an understanding of its dependencies and revisit the agreement when something changes it.
A promise deserves trust when it makes clear what we are accepting today and what commitment we are making for tomorrow. That clarity gives imagination a shared place from which to build.