Use AI without weakening client confidentiality, adviser responsibility or operational control.
Cytria helps independent asset managers identify current AI exposure, define governance controls and test one internal workflow without automating investment advice.
No client portfolio or confidential documents required for the first conversation.
Current exposure patterns in wealth management
Advisers and analysts often seek to increase productivity by using public tools to synthesize market analyses or read portfolio structures. Lacking formal rules, client data can transit through external servers, presenting direct regulatory challenges under FINMA circulars and nLPD constraints.
Confidentiality and governance risks
- FINMA Outsourcing Compliance: Unchecked cloud integrations fail client-data outsourcing requirements.
- Suitability & Advisory Risks: Hallucinated research could lead to inappropriate advice.
- No Verification Logs: Absence of audit trails for generated investment memos.
Best first internal workflows
Start with secure, internal workflows that augment RM efficiency without automating decision-making or portfolio adjustments.
1. Relationship-Manager Briefing
Input: Approved internal macro notes and research sheets.
Assisted Action: Draft client-meeting briefing bullet points.
Human Review: RM reviews and validates accuracy before client meeting.
Source / Log: Citations map back directly to internal research publication date.
Prohibited: No autonomous advisory emails or portfolio pitches.
2. Policy & Product Knowledge Search
Input: Compliance handbooks and approved fund factsheets.
Assisted Action: Retrieve investment restrictions and tax status.
Human Review: Compliance officer confirms product constraints before trade booking.
Source / Log: Verbatim extracts shown alongside source document revision date.
Prohibited: No automatic approvals of non-compliant assets.
3. Investment-Committee Preparation
Input: Raw market indicator logs and broker thematic notes.
Assisted Action: Synthesize draft agendas and investment arguments.
Human Review: Committee members review, edit, and vote on strategy.
Source / Log: Audit logs document sources and draft changes.
Prohibited: No automated trade orders or portfolio rebalancings.
4. Research Summarisation
Input: External broker research papers and macroeconomic files.
Assisted Action: Extract core changes in holding allocations.
Human Review: Portfolio manager reads summary and cross-references original papers.
Source / Log: PDF page numbers cited next to every statement.
Prohibited: No direct adjustments to the model portfolio without approval.
5. Onboarding Completeness Check
Input: ID files, self-declaration forms, and KYC profiles.
Assisted Action: Detect missing fields, signature gaps, or mismatching addresses.
Human Review: KYC specialist verifies gaps and approves client onboarding status.
Source / Log: Discrepancy report logged against check list.
Prohibited: No autonomous onboarding approvals or FINMA registry submissions.
6. Operational Exception Routing
Input: Settlement discrepancy logs and broker exception tickets.
Assisted Action: Classify ticket severity and suggest the operations owner.
Human Review: Ops manager checks routing and assigns the ticket.
Source / Log: Exception flags stored in transaction database logs.
Prohibited: No autonomous trade corrections or execution reversals.
What must remain under adviser responsibility
Advisers must retain final decision-making power. AI is restricted to summarising and draft support.
Exclusions: Cytria does not build autonomous investment recommenders or remove human compliance reviews.
Governance & Readiness Audit
We perform an on-site audit to establish your current AI exposure, check vendor terms, and outline a custom deployment roadmap.
View sample governance dossier →Swiss-hosted and client-controlled deployment options
We design and deploy open LLMs within your isolated Swiss VPC or on physical hardware. All files and model parameters remain completely isolated from public providers.
Founder & Managing Director
Led by Enrico Bartolotti, Cytria Sàrl provides independent wealth managers with sovereign AI governance. We operate with strict client data isolation guarantees.
Cytria Sàrl · Geneva CHE-378.854.076
Read about our team on the About Page.
Human Validation Model
All workflows force structured citation verification and multi-step manual validation blocks before execution.
Read our method →Architecture & Security
Zero multi-tenant sharing. Models run on dedicated virtual infrastructure protected by corporate firewalls.
Review security architecture →Ownership & Handover
All code, model integrations, and database schemas are handed over to you upon completion. Full reversibility.
Read engagement terms →Request a confidential review of your AI exposure
We will discuss your current processes, compliance constraints, and potential safe workflows. No client portfolios or files are required for this first conversation.