Trust, built in

Every profile CIS builds carries a confidence score for how much can be trusted about that person — and that score has a hard ceiling, well short of full marks, no matter how much is known. It's a deliberate admission that no system can fully know a human being, built into the maths rather than left as a disclaimer nobody reads.

How much you actually trust someone is set by you, the person who knows them — never estimated, guessed or overridden by AI. It sits apart from everything CIS scores automatically, on purpose.

Before CIS will suggest introducing someone, they have to clear an impact gate: are they credible, and do they leave things better than they found them. Spending more money cannot move that score. It's judged on conduct, not budget.

Whenever someone actually working a record spots it's wrong, it gets corrected there and then, as part of the work — not flagged for a clean-up that never comes. The more a record is genuinely worked, the more accurate it gets.

Before any introduction happens, the person making it has to agree to make it, and John has to sign it off. Nobody is added to an outreach list, or messaged, purely because CIS thought they'd be a good match — there's no automated send, ever. A real person always writes the note, and that same person always decides whether to send it. See the partner guide for exactly how an introduction works, step by step, and Your data & privacy for what CIS holds about someone before that point.

CIS is built to be one system for someone's whole working life, not just a sales pipeline — that ambition is real, and it's also still mostly proven on one pilot partner rather than at scale. The five things above are live today; the wider vision is where this is heading.