A forecast: if interfaces become conversational, familiar IT services will disappear from the surface

This is a postdiction - a reflection on what will happen if several trends already visible today continue in the same direction. I am writing this deliberately not as speculation, but as a practical question for people making architectural decisions right now.

Suppose that in three years the interfaces to IT systems - corporate, consumer, operational - become predominantly conversational. Not in the sense of "voice input instead of a keyboard." In the sense that the user states an intention in plain language, the system understands the context, and the action is performed. The earlier pieces on why voice interfaces matter beyond the phone and on what the Watson lesson taught us about business buying faster answers are the building blocks behind this scenario.

What happens then to the systems we are building today?

The interface will stop being a competitive advantage

A large part of the value of enterprise software today is UX: clean screens, sensible forms, well-organized menus. If the interface becomes conversational, that layer disappears. The user does not see how the screen is structured - they simply say what they want.

This means that companies competing at the interface level will lose that advantage almost entirely. What remains is what is underneath the interface: data, logic, processes, integrations.

Systems with internal coherence will win

A conversational interface is not a magic button. For it to work, the system must be able to interpret an intention and then find the right data or trigger the right process. For that to be possible, knowledge, data, and processes must be connected inside the system - not "in the heads of a few specialists," but formally, in structure.

This means: systems where data is scattered across spreadsheets, logic lives in email threads, and processes are documented nowhere except printed manuals - those systems will get nothing from a conversational interface. There is nowhere to get the answer from.

Internal search and navigation will change most dramatically

Today an employee who needs information goes to the intranet, to a folder on the server, or to a colleague. This is slow and depends on whether the person knows where to look.

If search becomes conversational, it can be as fast as a query against a well-organized knowledge base. But only if that knowledge base exists. If corporate knowledge is unstructured, conversational search will return either noise or a confidently delivered wrong answer.

Companies that start structuring their knowledge today - not for any specific product, but as an operational practice - will be in a much stronger position.

Which systems will have the hardest transition

The most difficult position will be for systems that:

• store logic in the interface rather than in data - for example, "the correct order to click the buttons" is documented nowhere except the screen itself;
• operate through highly specific data structures that the system understands but a person does not;
• integrate through file exchanges and manual operations rather than explicit APIs;
• have no description of their processes in either machine-readable or human-readable form.

That description fits the majority of enterprise IT systems installed before 2010.

What is worth doing now

There is no need to wait for conversational interfaces to start preparing. The following practices are valuable regardless of which direction interfaces go:

• document business logic explicitly, not only in code or in employees' heads;
• build APIs between systems instead of file exchanges;
• structure corporate knowledge so it can be used by someone other than the person who created it;
• when choosing new systems, look not only at the interface but at data quality and API openness.

If conversational interfaces arrive quickly - you will be ready. If they arrive slowly - you still win: from cleaner architecture, better-organized knowledge, and more manageable processes.