Productivity Software Is Dead. Long Live Context.

In 2000, Salesforce printed No Software inside a red circle with a line through it and built a company on the joke. The claim was never that software had stopped mattering. It was that one delivery model — install it, license it, maintain it — had reached the end of its useful life, and that everything resting on that assumption was about to look old.

Something similar is happening now, one layer up. The delivery model is fine. The organizing assumption underneath it is not.

Productivity software rests on a single premise: the bottleneck is execution. Give a person a better tool for the task in front of them — a faster spreadsheet, a cleaner CRM, a project tracker, a note-taker — and more work comes out the other end. That premise held for a long time, and it produced genuinely great software.

It also produced a specific pathology. Every new task got a new tool, and every tool became one more place where the story of the work lived, partially.

Workers toggle between applications roughly 1,200 times a day — close to four hours a week in pure reorientation, about 9% of the working year, spent before any of the actual work begins.

Usage-log study, three Fortune 500 companies

That is not a tool problem waiting for a better tool. It is the accumulated cost of an assumption.

What the constraint actually was

Watch where the week goes and a different bottleneck appears. Atlassian’s 2024 survey of 5,000 knowledge workers found 55% struggle to track down information even when they know exactly who has it, and half had discovered mid-project that another team inside the same company was already doing the same work. Microsoft’s 2025 Work Trend Index — built on product telemetry rather than self-report — puts the average worker at 117 emails and 153 messages a day, interrupted roughly every two minutes during core hours.

None of that describes a shortage of execution capacity. It describes a shortage of context: what happened before this, what got decided, what is still open, who already knows.

And supplying that context has never actually been anyone’s job. It happens in hallway conversations and direct messages, or it doesn’t happen at all. When someone leaves, it leaves with them. When work moves from one person to the next, it gets re-explained from the beginning — each retelling shorter and lossier than the one before it.

Why the assumption is only breaking now

Because until recently, nothing else was possible. Context could not be captured at the speed work happens. Transcription was expensive and unreliable; document understanding barely existed. The only available mechanism was a person choosing to write something down for someone else’s benefit later — and that mechanism competes directly with the work itself, so it loses.

That constraint is gone. Transcription and document processing are now close to free. Capture is a solved problem.

Curation is not. Deciding what matters, who needs it, and when is the entire remaining problem, and it is not a productivity problem. It does not get better by handing one person a faster tool. It gets solved when the work carries its own context forward.

What replaces it

Every era of enterprise software eventually produced a system of record. ERP for what a company owns. CRM for who it sells to. HRIS for who works there. Each started as a bundle of features and ended as the place a particular kind of truth lived.

The age of AI has an obvious missing record, and it is not another dashboard. It is the shared context that people — and now the models working beside them — need in order to understand how the work actually connects.

Productivity software is not dead because it failed. It is dead because it answered the question that was answerable at the time — make each task faster — and the question that remains is a different one entirely.

Long live context.

Written by Oded Noy, ContextApp.