Wealth-Management AI Use-Case Map

Each card identifies the user, problem, data, expected qualitative value, risks, human oversight, reusable components and a bounded first experiment.

FINMA’s 2025 survey shows broad AI activity across Swiss financial institutions alongside continuing work on strategy, governance, data and third-party risk.[FINMA, 24 April 2025]

47of 47 use cases

Qualitative value only · no invented ROI

Agentic AIRelationship management

Automated meeting preparation

Primary user

Relationship manager

Expected value

More preparation time, faster follow-up and more consistent service

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for automated meeting preparation.
AI pattern: Agentic AI
Required data: CRM, portfolio, approved research and interaction history
Main risks: Unsuitable personalisation, stale information and client confidentiality
Human oversight: Relationship manager reviews every client-facing output
Reusable components: Client context, retrieval, drafting, approval and CRM connectors
Suggested first experiment: Run a bounded four-to-six-week test for automated meeting preparation with one user group, approved data and a manual review step.

Generative AIRelationship management

Client portfolio briefing

Primary user

Relationship manager

Expected value

More preparation time, faster follow-up and more consistent service

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for client portfolio briefing.
AI pattern: Generative AI
Required data: CRM, portfolio, approved research and interaction history
Main risks: Unsuitable personalisation, stale information and client confidentiality
Human oversight: Relationship manager reviews every client-facing output
Reusable components: Client context, retrieval, drafting, approval and CRM connectors
Suggested first experiment: Run a bounded four-to-six-week test for client portfolio briefing with one user group, approved data and a manual review step.

Generative AIRelationship management

Meeting-note structuring

Primary user

Relationship manager

Expected value

More preparation time, faster follow-up and more consistent service

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for meeting-note structuring.
AI pattern: Generative AI
Required data: CRM, portfolio, approved research and interaction history
Main risks: Unsuitable personalisation, stale information and client confidentiality
Human oversight: Relationship manager reviews every client-facing output
Reusable components: Client context, retrieval, drafting, approval and CRM connectors
Suggested first experiment: Run a bounded four-to-six-week test for meeting-note structuring with one user group, approved data and a manual review step.

Agentic AIRelationship management

Follow-up action generation

Primary user

Relationship manager

Expected value

More preparation time, faster follow-up and more consistent service

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for follow-up action generation.
AI pattern: Agentic AI
Required data: CRM, portfolio, approved research and interaction history
Main risks: Unsuitable personalisation, stale information and client confidentiality
Human oversight: Relationship manager reviews every client-facing output
Reusable components: Client context, retrieval, drafting, approval and CRM connectors
Suggested first experiment: Run a bounded four-to-six-week test for follow-up action generation with one user group, approved data and a manual review step.

Predictive AIRelationship management

Relationship-manager knowledge assistant

Primary user

Relationship manager

Expected value

More preparation time, faster follow-up and more consistent service

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for relationship-manager knowledge assistant.
AI pattern: Predictive AI
Required data: CRM, portfolio, approved research and interaction history
Main risks: Unsuitable personalisation, stale information and client confidentiality
Human oversight: Relationship manager reviews every client-facing output
Reusable components: Client context, retrieval, drafting, approval and CRM connectors
Suggested first experiment: Run a bounded four-to-six-week test for relationship-manager knowledge assistant with one user group, approved data and a manual review step.

Generative AIRelationship management

Personalised communication drafting

Primary user

Relationship manager

Expected value

More preparation time, faster follow-up and more consistent service

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for personalised communication drafting.
AI pattern: Generative AI
Required data: CRM, portfolio, approved research and interaction history
Main risks: Unsuitable personalisation, stale information and client confidentiality
Human oversight: Relationship manager reviews every client-facing output
Reusable components: Client context, retrieval, drafting, approval and CRM connectors
Suggested first experiment: Run a bounded four-to-six-week test for personalised communication drafting with one user group, approved data and a manual review step.

Generative AIRelationship management

Prospect research

Primary user

Relationship manager

Expected value

More preparation time, faster follow-up and more consistent service

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for prospect research.
AI pattern: Generative AI
Required data: CRM, portfolio, approved research and interaction history
Main risks: Unsuitable personalisation, stale information and client confidentiality
Human oversight: Relationship manager reviews every client-facing output
Reusable components: Client context, retrieval, drafting, approval and CRM connectors
Suggested first experiment: Run a bounded four-to-six-week test for prospect research with one user group, approved data and a manual review step.

Predictive AIRelationship management

Client-event detection

Primary user

Relationship manager

Expected value

More preparation time, faster follow-up and more consistent service

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for client-event detection.
AI pattern: Predictive AI
Required data: CRM, portfolio, approved research and interaction history
Main risks: Unsuitable personalisation, stale information and client confidentiality
Human oversight: Relationship manager reviews every client-facing output
Reusable components: Client context, retrieval, drafting, approval and CRM connectors
Suggested first experiment: Run a bounded four-to-six-week test for client-event detection with one user group, approved data and a manual review step.

Agentic AIRelationship management

Next-best-conversation support

Primary user

Relationship manager

Expected value

More preparation time, faster follow-up and more consistent service

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for next-best-conversation support.
AI pattern: Agentic AI
Required data: CRM, portfolio, approved research and interaction history
Main risks: Unsuitable personalisation, stale information and client confidentiality
Human oversight: Relationship manager reviews every client-facing output
Reusable components: Client context, retrieval, drafting, approval and CRM connectors
Suggested first experiment: Run a bounded four-to-six-week test for next-best-conversation support with one user group, approved data and a manual review step.

Predictive AIInvestment and asset management

Investment-research synthesis

Primary user

Investment professional

Expected value

Faster synthesis and more consistent access to investment knowledge

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for investment-research synthesis.
AI pattern: Predictive AI
Required data: Research, market data, product documents and approved internal views
Main risks: Hallucination, source provenance, timeliness and suitability boundaries
Human oversight: Investment professional validates evidence and final interpretation
Reusable components: Research retrieval, source citation, summarisation and alerts
Suggested first experiment: Run a bounded four-to-six-week test for investment-research synthesis with one user group, approved data and a manual review step.

Generative AIInvestment and asset management

Portfolio commentary

Primary user

Investment professional

Expected value

Faster synthesis and more consistent access to investment knowledge

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for portfolio commentary.
AI pattern: Generative AI
Required data: Research, market data, product documents and approved internal views
Main risks: Hallucination, source provenance, timeliness and suitability boundaries
Human oversight: Investment professional validates evidence and final interpretation
Reusable components: Research retrieval, source citation, summarisation and alerts
Suggested first experiment: Run a bounded four-to-six-week test for portfolio commentary with one user group, approved data and a manual review step.

Generative AIInvestment and asset management

Product and mandate knowledge assistant

Primary user

Investment professional

Expected value

Faster synthesis and more consistent access to investment knowledge

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for product and mandate knowledge assistant.
AI pattern: Generative AI
Required data: Research, market data, product documents and approved internal views
Main risks: Hallucination, source provenance, timeliness and suitability boundaries
Human oversight: Investment professional validates evidence and final interpretation
Reusable components: Research retrieval, source citation, summarisation and alerts
Suggested first experiment: Run a bounded four-to-six-week test for product and mandate knowledge assistant with one user group, approved data and a manual review step.

Agentic AIInvestment and asset management

Investment-committee preparation

Primary user

Investment professional

Expected value

Faster synthesis and more consistent access to investment knowledge

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for investment-committee preparation.
AI pattern: Agentic AI
Required data: Research, market data, product documents and approved internal views
Main risks: Hallucination, source provenance, timeliness and suitability boundaries
Human oversight: Investment professional validates evidence and final interpretation
Reusable components: Research retrieval, source citation, summarisation and alerts
Suggested first experiment: Run a bounded four-to-six-week test for investment-committee preparation with one user group, approved data and a manual review step.

Predictive AIInvestment and asset management

Market-event summarisation

Primary user

Investment professional

Expected value

Faster synthesis and more consistent access to investment knowledge

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for market-event summarisation.
AI pattern: Predictive AI
Required data: Research, market data, product documents and approved internal views
Main risks: Hallucination, source provenance, timeliness and suitability boundaries
Human oversight: Investment professional validates evidence and final interpretation
Reusable components: Research retrieval, source citation, summarisation and alerts
Suggested first experiment: Run a bounded four-to-six-week test for market-event summarisation with one user group, approved data and a manual review step.

Agentic AIInvestment and asset management

Portfolio-monitoring alerts

Primary user

Investment professional

Expected value

Faster synthesis and more consistent access to investment knowledge

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for portfolio-monitoring alerts.
AI pattern: Agentic AI
Required data: Research, market data, product documents and approved internal views
Main risks: Hallucination, source provenance, timeliness and suitability boundaries
Human oversight: Investment professional validates evidence and final interpretation
Reusable components: Research retrieval, source citation, summarisation and alerts
Suggested first experiment: Run a bounded four-to-six-week test for portfolio-monitoring alerts with one user group, approved data and a manual review step.

Generative AIInvestment and asset management

Document and earnings-call analysis

Primary user

Investment professional

Expected value

Faster synthesis and more consistent access to investment knowledge

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for document and earnings-call analysis.
AI pattern: Generative AI
Required data: Research, market data, product documents and approved internal views
Main risks: Hallucination, source provenance, timeliness and suitability boundaries
Human oversight: Investment professional validates evidence and final interpretation
Reusable components: Research retrieval, source citation, summarisation and alerts
Suggested first experiment: Run a bounded four-to-six-week test for document and earnings-call analysis with one user group, approved data and a manual review step.

Generative AIInvestment and asset management

Research-library search

Primary user

Investment professional

Expected value

Faster synthesis and more consistent access to investment knowledge

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for research-library search.
AI pattern: Generative AI
Required data: Research, market data, product documents and approved internal views
Main risks: Hallucination, source provenance, timeliness and suitability boundaries
Human oversight: Investment professional validates evidence and final interpretation
Reusable components: Research retrieval, source citation, summarisation and alerts
Suggested first experiment: Run a bounded four-to-six-week test for research-library search with one user group, approved data and a manual review step.

Agentic AICompliance and risk

KYC file preparation

Primary user

Compliance or risk specialist

Expected value

Better preparation, consistency and traceability in review work

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for kyc file preparation.
AI pattern: Agentic AI
Required data: Policies, case files, client documents and control evidence
Main risks: False negatives, bias, explainability and inappropriate automation
Human oversight: Specialist owns conclusions, escalation and regulatory judgement
Reusable components: Document intake, policy retrieval, evidence trace and review queues
Suggested first experiment: Run a bounded four-to-six-week test for kyc file preparation with one user group, approved data and a manual review step.

Generative AICompliance and risk

Client-document review

Primary user

Compliance or risk specialist

Expected value

Better preparation, consistency and traceability in review work

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for client-document review.
AI pattern: Generative AI
Required data: Policies, case files, client documents and control evidence
Main risks: False negatives, bias, explainability and inappropriate automation
Human oversight: Specialist owns conclusions, escalation and regulatory judgement
Reusable components: Document intake, policy retrieval, evidence trace and review queues
Suggested first experiment: Run a bounded four-to-six-week test for client-document review with one user group, approved data and a manual review step.

Generative AICompliance and risk

Policy assistant

Primary user

Compliance or risk specialist

Expected value

Better preparation, consistency and traceability in review work

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for policy assistant.
AI pattern: Generative AI
Required data: Policies, case files, client documents and control evidence
Main risks: False negatives, bias, explainability and inappropriate automation
Human oversight: Specialist owns conclusions, escalation and regulatory judgement
Reusable components: Document intake, policy retrieval, evidence trace and review queues
Suggested first experiment: Run a bounded four-to-six-week test for policy assistant with one user group, approved data and a manual review step.

Generative AICompliance and risk

Regulation-change summarisation

Primary user

Compliance or risk specialist

Expected value

Better preparation, consistency and traceability in review work

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for regulation-change summarisation.
AI pattern: Generative AI
Required data: Policies, case files, client documents and control evidence
Main risks: False negatives, bias, explainability and inappropriate automation
Human oversight: Specialist owns conclusions, escalation and regulatory judgement
Reusable components: Document intake, policy retrieval, evidence trace and review queues
Suggested first experiment: Run a bounded four-to-six-week test for regulation-change summarisation with one user group, approved data and a manual review step.

Agentic AICompliance and risk

Compliance monitoring

Primary user

Compliance or risk specialist

Expected value

Better preparation, consistency and traceability in review work

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for compliance monitoring.
AI pattern: Agentic AI
Required data: Policies, case files, client documents and control evidence
Main risks: False negatives, bias, explainability and inappropriate automation
Human oversight: Specialist owns conclusions, escalation and regulatory judgement
Reusable components: Document intake, policy retrieval, evidence trace and review queues
Suggested first experiment: Run a bounded four-to-six-week test for compliance monitoring with one user group, approved data and a manual review step.

Agentic AICompliance and risk

Control-evidence preparation

Primary user

Compliance or risk specialist

Expected value

Better preparation, consistency and traceability in review work

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for control-evidence preparation.
AI pattern: Agentic AI
Required data: Policies, case files, client documents and control evidence
Main risks: False negatives, bias, explainability and inappropriate automation
Human oversight: Specialist owns conclusions, escalation and regulatory judgement
Reusable components: Document intake, policy retrieval, evidence trace and review queues
Suggested first experiment: Run a bounded four-to-six-week test for control-evidence preparation with one user group, approved data and a manual review step.

Agentic AICompliance and risk

Investigation support

Primary user

Compliance or risk specialist

Expected value

Better preparation, consistency and traceability in review work

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for investigation support.
AI pattern: Agentic AI
Required data: Policies, case files, client documents and control evidence
Main risks: False negatives, bias, explainability and inappropriate automation
Human oversight: Specialist owns conclusions, escalation and regulatory judgement
Reusable components: Document intake, policy retrieval, evidence trace and review queues
Suggested first experiment: Run a bounded four-to-six-week test for investigation support with one user group, approved data and a manual review step.

Agentic AICompliance and risk

Human-in-the-loop review agents

Primary user

Compliance or risk specialist

Expected value

Better preparation, consistency and traceability in review work

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for human-in-the-loop review agents.
AI pattern: Agentic AI
Required data: Policies, case files, client documents and control evidence
Main risks: False negatives, bias, explainability and inappropriate automation
Human oversight: Specialist owns conclusions, escalation and regulatory judgement
Reusable components: Document intake, policy retrieval, evidence trace and review queues
Suggested first experiment: Run a bounded four-to-six-week test for human-in-the-loop review agents with one user group, approved data and a manual review step.

Predictive AIBanking operations

Document intake and classification

Primary user

Operations specialist

Expected value

Lower handling time, clearer queues and faster exception resolution

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for document intake and classification.
AI pattern: Predictive AI
Required data: Cases, emails, documents, transactions and operating procedures
Main risks: Incorrect routing, unauthorised action and incomplete context
Human oversight: Operations staff approve sensitive or irreversible actions
Reusable components: Classification, workflow orchestration, case context and audit trail
Suggested first experiment: Run a bounded four-to-six-week test for document intake and classification with one user group, approved data and a manual review step.

Agentic AIBanking operations

Case and email triage

Primary user

Operations specialist

Expected value

Lower handling time, clearer queues and faster exception resolution

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for case and email triage.
AI pattern: Agentic AI
Required data: Cases, emails, documents, transactions and operating procedures
Main risks: Incorrect routing, unauthorised action and incomplete context
Human oversight: Operations staff approve sensitive or irreversible actions
Reusable components: Classification, workflow orchestration, case context and audit trail
Suggested first experiment: Run a bounded four-to-six-week test for case and email triage with one user group, approved data and a manual review step.

Agentic AIBanking operations

Payment-workflow support

Primary user

Operations specialist

Expected value

Lower handling time, clearer queues and faster exception resolution

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for payment-workflow support.
AI pattern: Agentic AI
Required data: Cases, emails, documents, transactions and operating procedures
Main risks: Incorrect routing, unauthorised action and incomplete context
Human oversight: Operations staff approve sensitive or irreversible actions
Reusable components: Classification, workflow orchestration, case context and audit trail
Suggested first experiment: Run a bounded four-to-six-week test for payment-workflow support with one user group, approved data and a manual review step.

Generative AIBanking operations

Exception investigation

Primary user

Operations specialist

Expected value

Lower handling time, clearer queues and faster exception resolution

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for exception investigation.
AI pattern: Generative AI
Required data: Cases, emails, documents, transactions and operating procedures
Main risks: Incorrect routing, unauthorised action and incomplete context
Human oversight: Operations staff approve sensitive or irreversible actions
Reusable components: Classification, workflow orchestration, case context and audit trail
Suggested first experiment: Run a bounded four-to-six-week test for exception investigation with one user group, approved data and a manual review step.

Predictive AIBanking operations

Reconciliation assistance

Primary user

Operations specialist

Expected value

Lower handling time, clearer queues and faster exception resolution

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for reconciliation assistance.
AI pattern: Predictive AI
Required data: Cases, emails, documents, transactions and operating procedures
Main risks: Incorrect routing, unauthorised action and incomplete context
Human oversight: Operations staff approve sensitive or irreversible actions
Reusable components: Classification, workflow orchestration, case context and audit trail
Suggested first experiment: Run a bounded four-to-six-week test for reconciliation assistance with one user group, approved data and a manual review step.

Agentic AIBanking operations

Corporate-action support

Primary user

Operations specialist

Expected value

Lower handling time, clearer queues and faster exception resolution

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for corporate-action support.
AI pattern: Agentic AI
Required data: Cases, emails, documents, transactions and operating procedures
Main risks: Incorrect routing, unauthorised action and incomplete context
Human oversight: Operations staff approve sensitive or irreversible actions
Reusable components: Classification, workflow orchestration, case context and audit trail
Suggested first experiment: Run a bounded four-to-six-week test for corporate-action support with one user group, approved data and a manual review step.

Generative AIBanking operations

Operational knowledge assistant

Primary user

Operations specialist

Expected value

Lower handling time, clearer queues and faster exception resolution

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for operational knowledge assistant.
AI pattern: Generative AI
Required data: Cases, emails, documents, transactions and operating procedures
Main risks: Incorrect routing, unauthorised action and incomplete context
Human oversight: Operations staff approve sensitive or irreversible actions
Reusable components: Classification, workflow orchestration, case context and audit trail
Suggested first experiment: Run a bounded four-to-six-week test for operational knowledge assistant with one user group, approved data and a manual review step.

Generative AIBanking operations

Incident analysis

Primary user

Operations specialist

Expected value

Lower handling time, clearer queues and faster exception resolution

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for incident analysis.
AI pattern: Generative AI
Required data: Cases, emails, documents, transactions and operating procedures
Main risks: Incorrect routing, unauthorised action and incomplete context
Human oversight: Operations staff approve sensitive or irreversible actions
Reusable components: Classification, workflow orchestration, case context and audit trail
Suggested first experiment: Run a bounded four-to-six-week test for incident analysis with one user group, approved data and a manual review step.

Agentic AICOO and management

Management-report preparation

Primary user

COO or management team

Expected value

Shorter reporting cycles and more time for management judgement

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for management-report preparation.
AI pattern: Agentic AI
Required data: Management information, KPIs, plans, risks and financial data
Main risks: Misleading aggregation, weak lineage and overconfident commentary
Human oversight: Accountable owner validates numbers, narrative and decisions
Reusable components: Data lineage, analytics, narrative generation and exception alerts
Suggested first experiment: Run a bounded four-to-six-week test for management-report preparation with one user group, approved data and a manual review step.

Generative AICOO and management

KPI commentary

Primary user

COO or management team

Expected value

Shorter reporting cycles and more time for management judgement

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for kpi commentary.
AI pattern: Generative AI
Required data: Management information, KPIs, plans, risks and financial data
Main risks: Misleading aggregation, weak lineage and overconfident commentary
Human oversight: Accountable owner validates numbers, narrative and decisions
Reusable components: Data lineage, analytics, narrative generation and exception alerts
Suggested first experiment: Run a bounded four-to-six-week test for kpi commentary with one user group, approved data and a manual review step.

Predictive AICOO and management

Risk and anomaly alerts

Primary user

COO or management team

Expected value

Shorter reporting cycles and more time for management judgement

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for risk and anomaly alerts.
AI pattern: Predictive AI
Required data: Management information, KPIs, plans, risks and financial data
Main risks: Misleading aggregation, weak lineage and overconfident commentary
Human oversight: Accountable owner validates numbers, narrative and decisions
Reusable components: Data lineage, analytics, narrative generation and exception alerts
Suggested first experiment: Run a bounded four-to-six-week test for risk and anomaly alerts with one user group, approved data and a manual review step.

Generative AICOO and management

Executive briefings

Primary user

COO or management team

Expected value

Shorter reporting cycles and more time for management judgement

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for executive briefings.
AI pattern: Generative AI
Required data: Management information, KPIs, plans, risks and financial data
Main risks: Misleading aggregation, weak lineage and overconfident commentary
Human oversight: Accountable owner validates numbers, narrative and decisions
Reusable components: Data lineage, analytics, narrative generation and exception alerts
Suggested first experiment: Run a bounded four-to-six-week test for executive briefings with one user group, approved data and a manual review step.

Predictive AICOO and management

Project portfolio reporting

Primary user

COO or management team

Expected value

Shorter reporting cycles and more time for management judgement

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for project portfolio reporting.
AI pattern: Predictive AI
Required data: Management information, KPIs, plans, risks and financial data
Main risks: Misleading aggregation, weak lineage and overconfident commentary
Human oversight: Accountable owner validates numbers, narrative and decisions
Reusable components: Data lineage, analytics, narrative generation and exception alerts
Suggested first experiment: Run a bounded four-to-six-week test for project portfolio reporting with one user group, approved data and a manual review step.

Agentic AICOO and management

Budget and forecast support

Primary user

COO or management team

Expected value

Shorter reporting cycles and more time for management judgement

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for budget and forecast support.
AI pattern: Agentic AI
Required data: Management information, KPIs, plans, risks and financial data
Main risks: Misleading aggregation, weak lineage and overconfident commentary
Human oversight: Accountable owner validates numbers, narrative and decisions
Reusable components: Data lineage, analytics, narrative generation and exception alerts
Suggested first experiment: Run a bounded four-to-six-week test for budget and forecast support with one user group, approved data and a manual review step.

Generative AICOO and management

Process-mining insights

Primary user

COO or management team

Expected value

Shorter reporting cycles and more time for management judgement

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for process-mining insights.
AI pattern: Generative AI
Required data: Management information, KPIs, plans, risks and financial data
Main risks: Misleading aggregation, weak lineage and overconfident commentary
Human oversight: Accountable owner validates numbers, narrative and decisions
Reusable components: Data lineage, analytics, narrative generation and exception alerts
Suggested first experiment: Run a bounded four-to-six-week test for process-mining insights with one user group, approved data and a manual review step.

Predictive AITechnology

Software-development assistants

Primary user

Engineer or IT specialist

Expected value

Faster diagnosis, better documentation and improved delivery flow

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for software-development assistants.
AI pattern: Predictive AI
Required data: Code, tests, tickets, incidents, runbooks and technical documentation
Main risks: Insecure code, data exposure, licensing and unreliable remediation
Human oversight: Technical owner reviews code, changes and production actions
Reusable components: Code context, retrieval, test runners, ticketing and observability
Suggested first experiment: Run a bounded four-to-six-week test for software-development assistants with one user group, approved data and a manual review step.

Agentic AITechnology

Automated testing

Primary user

Engineer or IT specialist

Expected value

Faster diagnosis, better documentation and improved delivery flow

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for automated testing.
AI pattern: Agentic AI
Required data: Code, tests, tickets, incidents, runbooks and technical documentation
Main risks: Insecure code, data exposure, licensing and unreliable remediation
Human oversight: Technical owner reviews code, changes and production actions
Reusable components: Code context, retrieval, test runners, ticketing and observability
Suggested first experiment: Run a bounded four-to-six-week test for automated testing with one user group, approved data and a manual review step.

Generative AITechnology

Incident investigation

Primary user

Engineer or IT specialist

Expected value

Faster diagnosis, better documentation and improved delivery flow

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for incident investigation.
AI pattern: Generative AI
Required data: Code, tests, tickets, incidents, runbooks and technical documentation
Main risks: Insecure code, data exposure, licensing and unreliable remediation
Human oversight: Technical owner reviews code, changes and production actions
Reusable components: Code context, retrieval, test runners, ticketing and observability
Suggested first experiment: Run a bounded four-to-six-week test for incident investigation with one user group, approved data and a manual review step.

Generative AITechnology

Technical-documentation generation

Primary user

Engineer or IT specialist

Expected value

Faster diagnosis, better documentation and improved delivery flow

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for technical-documentation generation.
AI pattern: Generative AI
Required data: Code, tests, tickets, incidents, runbooks and technical documentation
Main risks: Insecure code, data exposure, licensing and unreliable remediation
Human oversight: Technical owner reviews code, changes and production actions
Reusable components: Code context, retrieval, test runners, ticketing and observability
Suggested first experiment: Run a bounded four-to-six-week test for technical-documentation generation with one user group, approved data and a manual review step.

Predictive AITechnology

Legacy-code understanding

Primary user

Engineer or IT specialist

Expected value

Faster diagnosis, better documentation and improved delivery flow

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for legacy-code understanding.
AI pattern: Predictive AI
Required data: Code, tests, tickets, incidents, runbooks and technical documentation
Main risks: Insecure code, data exposure, licensing and unreliable remediation
Human oversight: Technical owner reviews code, changes and production actions
Reusable components: Code context, retrieval, test runners, ticketing and observability
Suggested first experiment: Run a bounded four-to-six-week test for legacy-code understanding with one user group, approved data and a manual review step.

Agentic AITechnology

Support-ticket analysis

Primary user

Engineer or IT specialist

Expected value

Faster diagnosis, better documentation and improved delivery flow

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for support-ticket analysis.
AI pattern: Agentic AI
Required data: Code, tests, tickets, incidents, runbooks and technical documentation
Main risks: Insecure code, data exposure, licensing and unreliable remediation
Human oversight: Technical owner reviews code, changes and production actions
Reusable components: Code context, retrieval, test runners, ticketing and observability
Suggested first experiment: Run a bounded four-to-six-week test for support-ticket analysis with one user group, approved data and a manual review step.

Generative AITechnology

Knowledge retrieval for IT teams

Primary user

Engineer or IT specialist

Expected value

Faster diagnosis, better documentation and improved delivery flow

View use-case brief

Problem: Time and attention are fragmented across the information and workflow needed for knowledge retrieval for it teams.
AI pattern: Generative AI
Required data: Code, tests, tickets, incidents, runbooks and technical documentation
Main risks: Insecure code, data exposure, licensing and unreliable remediation
Human oversight: Technical owner reviews code, changes and production actions
Reusable components: Code context, retrieval, test runners, ticketing and observability
Suggested first experiment: Run a bounded four-to-six-week test for knowledge retrieval for it teams with one user group, approved data and a manual review step.

Start with the workflow. Then choose the AI pattern.

Automated meeting preparation

Client portfolio briefing

Meeting-note structuring

Follow-up action generation

Relationship-manager knowledge assistant

Personalised communication drafting

Prospect research

Client-event detection

Next-best-conversation support

Investment-research synthesis

Portfolio commentary

Product and mandate knowledge assistant

Investment-committee preparation

Market-event summarisation

Portfolio-monitoring alerts

Document and earnings-call analysis

Research-library search

KYC file preparation

Client-document review

Policy assistant

Regulation-change summarisation

Compliance monitoring

Control-evidence preparation

Investigation support

Human-in-the-loop review agents

Document intake and classification

Case and email triage

Payment-workflow support

Exception investigation

Reconciliation assistance

Corporate-action support

Operational knowledge assistant

Incident analysis

Management-report preparation

KPI commentary

Risk and anomaly alerts

Executive briefings

Project portfolio reporting

Budget and forecast support

Process-mining insights

Software-development assistants

Automated testing

Incident investigation

Technical-documentation generation

Legacy-code understanding

Support-ticket analysis

Knowledge retrieval for IT teams