A quality bar for AI work.

Did the system perform the task that was actually requested?

A detailed answer to the wrong question still fails.

Did it find and use the information the method needs?

Strong reasoning cannot compensate for missing context.

Are important claims supported by evidence that can be inspected?

Is the evidence current enough for the claim being made?

Does the output make coverage limits visible instead of treating a recent check as complete knowledge?

Can the reader tell what was observed and what the system concluded from it?

Does the analysis reflect the right method for the task, or could the same generic prose have been produced for almost anything?

Did the system surface what matters?

A dump of everything found is not the same thing as judgment.

Can the intended person use the output in the real situation that created the need?

When the information is weak, does the system say so?

The right failure can be more useful than invented certainty.

Are the next steps proportionate to the evidence and the company context available?

We run the same work with different combinations of method and context to see where the quality changes.

One useful comparison is the AI tool alone, the AI tool with a skill, and the AI tool with both the skill and LMTY. That helps separate what the method adds from what maintained context and history add.

Did each run start from the same conditions?

We define the criteria before seeing the results, use the same question, model, and settings, and start each run in a fresh session so earlier outputs or saved memory do not affect the comparison.

We intentionally test conditions that make AI work look better than it is.

Stale sources. Conflicting evidence. Weak coverage. Marketing claims that sound like facts. One anecdote that looks like a trend. Company assumptions contradicted by customer evidence.

The final question is practical.

Would a strong practitioner trust this enough to use it?