Artificial intelligence has become a feature, a strategy, and a sentence added to investor presentations. Products that once promised speed or simplicity now promise intelligence. Sometimes the new capability is useful. Sometimes AI is present because its absence has become harder to explain than its purpose.
The question for a product team should not be whether AI can be added. It should be whether uncertainty, automation, and new costs improve the user’s actual task.
Start with the job, not the model
A useful product begins with a person trying to make progress. They may need to compare a large amount of information, draft a first version, detect an unusual pattern, or translate messy input into a workable structure. These are jobs where AI may reduce effort.
‘Chat with your data’ is not a user need. Neither is ‘AI-powered insights.’ Those phrases describe an interface and a technology. A team still has to name what becomes easier, for whom, and how success will be recognised.
Probabilistic features create new product debt
Traditional software can fail. AI features can also produce an answer that looks complete while being subtly wrong. That changes support, quality assurance, interface design, and the language used to set expectations.
If a user must verify every output, the product should be honest about that labour. A feature that saves five minutes but creates ten minutes of checking has moved effort rather than removed it.
Sometimes a rule is the smarter technology
Many product problems are predictable. A clear search filter, a well-written template, a deterministic recommendation, or one better default can outperform a model while remaining faster, cheaper, and easier to understand.
Choosing not to use AI is not a failure of ambition. It can be evidence that a team understands its problem well enough to use the simplest dependable tool.
A product does not become more intelligent when it contains AI. It becomes more intelligent when its technology fits the problem, its limits are visible, and the user leaves with less work than they brought in.
Back to all writing