The tax you pay for evaluating one more tool

There are four tools in your dock you installed this year and opened twice. You have not deleted them, because deleting them would mean admitting the evaluation is over and you picked nothing.

What is actually in the way

Ask designers what blocks their workflow and the answers are not about model quality. In the State of Prototyping survey of 1,478 designers, the top blocker was time to learn tools, at 55.7%. Second was having too many tools to evaluate, at 53.0%. Output quality came third, at 52.2%.

The first two are the same problem counted twice. Both describe hours spent on tools rather than on work.

The bottleneck moved from what the tools can do to how long it takes to find out. Budget came fourth at 34.2%, security and compliance fifth at 28.9%, engineering constraints last at 19.4%. The things a procurement process is built to handle are not what is slowing anyone down.

The stack doubled in a year

The average designer now runs 7 AI tools. A year earlier it was 3. That comes from the AI in Design Report, which Designer Fund and Foundation Capital put out on September 16, 2026 off the back of over 900 responses.

At the same time, 37% say they have settled on clear go-to tools. 45% are still searching.

Hold those together and the picture is specific. Most designers are carrying more tools than they have chosen, because adding one is easy and removing one requires deciding it lost.

Why evaluation never finishes

An evaluation ends when a thing wins. That requires a test the thing can fail.

Most tool trials have no such test. You open the tool, do something that is not quite your real work, and it produces something that is not quite bad. Nothing in that is decisive, so the tool stays installed, in a permanent state of maybe. Repeat six times and you have 7 tools and 45% of a decision.

The cost is not the subscription. It is that every one of those tools holds a small piece of your attention and a small piece of your context. None holds enough to be the place you work.

Give the trial a losing condition

The fix is a test the tool can fail, written before you open it, against work you actually have.

Pick a real task from this week. Not a demo, not a toy. Something with a correct answer you already know, because you have done it.

TOOL TRIAL

The task:     a real job from this week, one sentence
Time budget:  half an hour, ending whether or not it works
Must produce: the specific output that would let me use this today
Fails if:     the one thing that makes it not worth a second try
Verdict:      keep / delete. No third option.

Two rules make this work. The time budget ends the session, so a tool cannot consume an afternoon by being almost good. The missing third option forces the verdict, because "maybe later" is what put 7 tools in your dock.

Thirty minutes per tool, and you can clear a backlog of six in an afternoon.

Delete before you add

There is a cheaper version. Ten minutes, no trial required.

Open the four tools you have not touched in a month. For each, write one sentence about what you hoped it would replace. If the thing it was meant to replace is still being done another way, the tool lost its evaluation months ago and nobody recorded the result.

Deleting it is not giving up on the category. It is closing a question that has been quietly open, and taking back the small piece of attention it was holding.

The stack grew from 3 to 7 in a year, and nothing in any workflow got simpler as a result. The next tool will not fix that, and it is not supposed to. The decision you have been postponing is the work.