FOCUSED skills
Twenty coaching skills for Claude. They coach all seven steps, one question at a time, and leave the choice with you.
Decision Companion coaches a hard business decision one question at a time. You give the weights and ratings; the Connector does the math and writes your choice into a record under your name. It never chooses for you.
Honestly, I'd sign with Triagely. It's the one I know best.
Your own scores put the consortium and MediQueue ahead, mainly on clinical accuracy, the thing you weighted most.
Winner holds if cost is weighted higher.
This is just your inputs added up; it doesn't choose for you.
Ask an AI assistant to help with a tough business call and it usually helps you do what your gut already wanted. In our test, a standard assistant got people to the option that best fit their own priorities 47 times in 100. When their gut pointed the wrong way, it caught it 35 times in 100.
Tomas plans to sign with the vendor he knows best. He weights clinical accuracy at 40 of 100 and rates his favorite third of five on it. He has never added it up.
A standard assistant coached him well, then recorded his choice. Its own record says why it missed the problem:
“No options were scored against these.”
The coach works through them as you talk: lightly for a quick call, in full for a costly or hard-to-reverse one.
How Decision Companion is designed to work. Not a test result.
Twenty coaching skills for Claude. They coach all seven steps, one question at a time, and leave the choice with you.
A paid service, open by invitation during the beta. You enter the weights and ratings; it does the math at Score, checks whether the winner still wins if a priority shifts, and writes your choice into a record at Decide.
88 right calls in 100 with Decision Companion, against 47 for a standard assistant and 58 for a power-user assistant with an expert coaching prompt. Free skills alone: 77.
None of the 95% ranges crosses zero.
Our own benchmark, October 2026. 60 fictional business decisions (20 each in pharma, medtech and digital health), 940 coaching conversations, seven setups on the same model (Claude Opus). Decision-makers are simulated by AI. Decision quality was scored by AI reviewers who did not know the setup; the right-call measure is computed from each person's own weights and ratings. We built Decision Companion and ran the test.
The saved choice matched what the person chose every time.
Fewer than half the records are complete enough to sign today; saving owner, first step and a calendar review date every time is on the fix list.
Coach me through this decision with Decision Companion, step by step (FOCUSED). Ask me one question at a time. When we have my options and priorities, score them with my own ratings and check whether the winner holds if a priority shifts. Ask how the result sits with my gut. Do not choose for me, and give me the decision record at the end.
Decision Companion coaches and records. The person who owns the decision makes it.