Agents & Applications

All topics / Software & Agents

For most of software history, an application has been a relatively fixed bundle of capabilities, presented through an interface designed for a human operator. Agents are changing both assumptions.

This collection starts with the relationship between users, their agents and individual applications, then explores four connected ideas. The arguments range from emerging interaction patterns we can already recognise to a longer-term possibility: software assembled, extended and sometimes created at the point of need, rather than purchased as a predefined bundle.

Start here

Nobody wants your in-app crapbot — Why the user's own agent may be better positioned to understand their goals than an assistant embedded in a single application, and why applications should expose tools to those agents.

Four connected arguments

1. Three Modes of Application Interaction distinguishes human-direct, agent-mediated and hybrid modes of interacting with applications.
2. The Disappearing Interface argues that product capability can grow while its human interface becomes simpler—and potentially more dynamic.
3. Applying Intelligence Late considers what changes when agents familiar with a user's needs discover, configure and compose software at the point of use.
4. The Disaggregation of Software explores how those changes may erode fixed application bundles, making room for narrow, specialist and newly constructed capabilities.

Each essay links to smaller articles explaining a single concept or a supporting argument. The paths suggest a reading order, but the ideas form a graph: the different 10% problem, for example, connects interface design to software slivers.

These are hypotheses about a transition, not claims that conventional applications will disappear on a particular date. Integrated products, stable interfaces and specialised engines still have substantial advantages. The interesting question is how far their boundaries will move.

Back to Software & Agents