Automation that removes friction, not just clicks

Friction tends to hide in the spaces between steps, not in the steps themselves. It lives in the handoff between teams — when a support request moves from the person who received it to the person who can resolve it, and critical context gets lost along the way. It lives in the wait — when someone needs a decision but the decision-maker does not know the request exists.

It lives in the search — when a team member spends twenty minutes looking for a document that should have been easy to find. It lives in the repeat — when the same question gets asked three times because the answer was never captured in a reusable place.

Mapping these friction points is the first step toward meaningful automation. The exercise is surprisingly simple: follow a single request or task from start to finish, and note every point where it slows down, waits, or depends on someone remembering something. The results often reveal that 80% of the elapsed time is spent waiting, not working.

Organizations that adopt the "thinnest viable platform" approach — creating systems with just enough capability to reduce friction without over-engineering — tend to achieve better outcomes than those that build elaborate automation before understanding where the real problems are.

Good automation removes those pressure points. It shortens the path, reduces ambiguity, and makes the next action easier. In that sense, the best automation feels less like a robot repeating work and more like a system removing unnecessary drag.

A practical example: instead of automating data entry from one system to another (which saves keystrokes), automate the notification that tells the right person what needs attention, with enough context for them to act immediately (which saves a full decision cycle). The first is click reduction. The second is friction reduction.

Another example: instead of automating report generation on a schedule (which produces documents no one reads), automate the surfacing of exceptions and changes that actually require human judgment. A weekly summary that says "everything is on track" is noise. An alert that says "project X is three days behind because the client has not responded to the design review" is actionable friction reduction.

The principle extends to AI-assisted workflows. An AI agent that drafts a client update saves time. An AI agent that drafts a client update, attaches the relevant files, identifies unresolved questions, and suggests who needs to review it before sending — that agent is removing friction from the workflow, not just automating a task.

Automation can create new friction when it is poorly designed. Excessive notifications, rigid approval chains that do not match how decisions actually get made, automated processes that break silently and produce incorrect outputs — these are all examples of automation that makes the system worse.

The most common anti-pattern is automating a broken process. If the underlying workflow is unclear, adding automation just makes it fail faster. The first step should always be understanding the process — who does what, why, and where it breaks — before introducing any automation. The rule is simple: fix the process, then automate it. Never automate a mess.