Why these systems run on Claude Opus 4.6

One model runs every ComplianceSME system: Claude Opus 4.6. This is not a preference. It is the licence condition, written into every file we ship, and the file checks for it before it does anything else.

AI models are not interchangeable. Each one follows instructions differently, and most treat written instructions as suggestions. A compliance record built that way is worthless on the day an auditor reads it. Before launch we benchmarked our systems across model tiers. Claude Opus 4.6 followed the written procedure to the letter on every run, produced the same result from the same inputs every time, and refused work outside its task. Other tiers skipped steps, reworded requirements, or answered questions they should have declined. So we licensed our systems to the one model that passed, and locked the rest out.

Every file states this licence in its opening rules. Load it on any other model and it declines to run, names the licence, and tells you which model to select. Load it on Claude Opus 4.6 and it runs the interview with you, one section at a time, and produces your compliance records.

Selecting the model takes one click. In Claude, the model menu sits at the top of every chat. Select Claude Opus 4.6 there before you type START. If you run our free systems inside a project named ComplianceSME Opus 4.6, the name on the door reminds you every time.

There are two ways to run these systems, and only one of them counts as a compliance record. Before you begin explains both, and the printed guide in your starter pack, Print me first, sets up the serious lane in ten minutes.