A sensitive client follow-up system for advisory teams

Consider a specialist advisory team of 18 people working with clients who bring sensitive personal, financial, or operational situations.

The team is not a clinical care provider, but the work still requires care. Clients may be under stress. Details may be confidential. Follow-up may depend on personal context, prior commitments, family dynamics, business pressure, or legal and financial constraints. A missed call or poorly prepared message can damage trust.

The team needs better infrastructure for follow-up. It does not need AI to decide what is best for the client.

This is where Proximity could be tailored as a sensitive client follow-up system.

The system's role is to preserve context, prepare review, and make escalation visible. It should not automate care, professional advice, emotional judgment, diagnosis, risk assessment, or client-facing decisions.

The Workflow

The workflow is weekly client follow-up review.

The team wants to know:

• Which clients are waiting for follow-up.
• What was promised and by whom.
• What sensitive context should shape the next contact.
• Which documents, notes, or prior decisions matter.
• Which situations need senior review before any message is sent.
• Which follow-ups are routine and which need escalation.

This workflow is common in relationship-heavy professional services: advisory, wealth, immigration, family office, specialist consulting, and other contexts where the client relationship depends on continuity and care.

The problem is not that professionals lack judgment. The problem is that the context around judgment is scattered across notes, calls, email, tasks, documents, and memory.

What Proximity Models

For this deployment, Proximity would model the client follow-up as the core work object.

Approved sources might include:

• Client notes.
• Meeting summaries.
• Email commitments.
• Task lists.
• Document requests.
• Prior decision logs.
• Internal review notes.
• Escalation rules.
• Consent or permission records where relevant.

The system would structure context around:

• Client, matter, owner, and relationship lead.
• Promise, source, due date, and responsible person.
• Sensitive context and access boundary.
• Missing document or unanswered question.
• Follow-up draft status.
• Required reviewer.
• Escalation reason.

The privacy examples in this article are broader than the advisory scenario. WHO guidance on AI for health stresses ethics, human rights, and responsible use. HIPAA in the United States is a concrete privacy regime for protected health information. NICE's evidence standards for digital health technologies also show that higher-impact systems require stronger evidence and governance. The general lesson applies beyond healthcare: sensitive client work needs context, privacy, evidence, and human accountability.

What The System Prepares

Before the weekly review, Proximity could prepare a client follow-up queue.

Each item might include:

• What was promised.
• Where the promise came from.
• Relevant client context.
• Missing documents or open questions.
• Suggested owner.
• Draft internal note or message for review.
• Escalation reason if the system detects sensitivity, uncertainty, or overdue follow-up.

The system could also prepare handover briefs when a relationship owner is absent or a client moves between teams. The brief should explain what changed, what was promised, which context matters, and what remains unresolved.

The point is not to create a more automated client relationship. The point is to make the human relationship less dependent on fragile memory.

What Remains Human

Care and professional judgment remain human.

Proximity must not diagnose, assess wellbeing, decide client needs, give professional advice, determine the correct emotional tone, send sensitive messages without review, or decide whether escalation is required in a final sense. It can flag patterns and prepare context, but a responsible person must review.

Humans remain responsible for:

• Professional advice.
• Client care judgment.
• Relationship judgment.
• Escalation decisions.
• Final message content.
• Privacy decisions.
• Sensitive disclosures.
• Any commitment made to the client.

The product should make this boundary explicit. Drafts should be drafts. Escalation queues should route to named humans. Sensitive context should respect permissions. The system should show source records and uncertainty instead of smoothing everything into confident language.

NIST's AI Risk Management Framework is relevant because it treats AI risk as a sociotechnical governance issue. For sensitive client work, the workflow must support accountable people, not hide responsibility behind automation.

Pilot Shape

A good pilot would focus on one client group and one follow-up rhythm.

For example, the team might choose clients in an active transition period where follow-up is frequent and context matters. The first phase would map the review cadence and source records. The second phase would create follow-up queues from approved notes, tasks, and commitments. The third phase would test handover briefs and escalation flags in weekly review.

Success signals include:

• Fewer missed follow-ups.
• Better prepared client calls.
• Clearer ownership of commitments.
• Faster handover when owners are unavailable.
• More consistent escalation of sensitive items for review.
• Less time spent searching for prior context.