Security and AI trust
The real engineering behind the AI
Succession Navigator is a growing platform built by a founder with a background in enterprise B2B software. Here's the real engineering behind the AI, and here's what's true today about security. We keep this page updated as the platform matures.
There is a lot of AI slop about — a chat box bolted onto a thin web wrapper, sold as intelligence. That scepticism is healthy, and this page exists to answer it with specifics rather than reassurances.
AI proposes. Humans decide.
Raw meeting transcripts are captured and locked — the AI is never allowed to edit or overwrite the original wording. Anything it infers or suggests, whether that’s a timeline, a family member’s intent or a possible risk, sits as a proposal only. It does not become part of the official case record until a human facilitator reviews the source and explicitly approves it.
Fact and assumption are structurally different things.
A participant’s statement, an assumption, the facilitator’s interpretation and a verified fact are stored as different kinds of thing, not as one undifferentiated pile of text. The AI cannot silently turn a passing comment into a binding family decision, and you get a defensible record of what was said, who said it, what supports it and what still needs testing.
Contradictions are preserved, not resolved by guesswork.
If a family member’s statement conflicts with something said earlier, the system flags the conflict and holds both statements. It does not pick one, average them, or let the later one quietly overwrite the earlier. Resolving it is the adviser’s job, done with the family, on the record.
A dedicated validator blocks unsafe definitive advice.
Text that reads as definitive legal, tax or financial advice — something like "this transfer will be tax-free" — is flagged and blocked from reaching any client-facing document until a licensed professional signs off. The system fails closed rather than letting confident-sounding AI text slip through to a family.
The family tree is built from real data, not AI guesswork.
The genogram is a deterministic diagram plotted from verified database relationships, not an AI-generated image. That is why it handles blended families, step-children and multiple generations with actual structural accuracy instead of a plausible-looking approximation.
Strict multi-tenant separation.
Every request is validated against the specific client and reseller organisation on the server. One family’s information can never surface in another family’s case, and staff only see the clients they are assigned to.
Internal notes stay internal by default.
Facilitator notes and unverified AI observations default to adviser-only visibility. Making something client-facing is a deliberate, separate action — never a side effect of writing it down.
In the product
Record changed or conflicting information transparently
Succession conversations evolve, and family members do not always agree. The misalignment workflow lets advisers link differing statements, classify the issue and document an appropriate resolution path without deleting either source. This creates a fairer, more auditable record and helps important disagreements remain visible until they are genuinely addressed.
Access control
Give the right people the right level of access
Succession Navigator separates practice membership from platform-wide authority and makes role management explicit. Practice Administrators can review users, assign appropriate roles and see whether each membership is active. This supports practical delegation while helping the organisation protect sensitive family and business information.
Professional boundaries
The platform stays in its lane
Succession Navigator supports your advisory process. It does not replace your professional judgement, and it does not issue legal, tax or financial advice. Where a decision depends on a discipline outside your own, the platform identifies that and prepares a redacted, professional-safe brief for the right specialist.
It also does not resolve family conflict. It surfaces and tracks disagreement professionally, and leaves the resolving to you.
Kept current
This page describes what is true today. As the platform matures, claims get added here only once they are confirmed — and this note stays until there is nothing left pending.
Bring the hard questions
Thirty minutes with the product open. Security and AI governance questions are welcome — they're the ones worth asking.
See the product, how it fits your practice, and how it could work for your clients. No obligation.