Built for SwissHacks 2026
Agent Angelo
The world's first proactive wealth management dashboard — an AI agent that doesn't wait to be asked.
agentangelo.financeWhat is Agent Angelo?
Agent Angelo is a next-generation Relationship Manager Intelligence Platform for Swiss and international private banks serving (ultra-)high-net-worth clients. It transforms years of CRM notes, live portfolio data, and real-time market signals into personalised, explainable, auditable advisory actions — while keeping the RM fully in the loop.
It is not just a dashboard. It is a living agent embedded in your data.
The idea of an agent living inside data — believing its world is real — was inspired by the art project PromptedWorld, where an AI agent exists entirely within a simulated reality, convinced that it and its world are genuine. Agent Angelo carries that same premise into finance: a mind living inside your data, acting as if the stakes are real — because they are.
Key Innovations
Proactive AI that calls you
Angelo monitors your client book 24/7 and calls you on your real phone when something urgent happens — in a human voice, with a two-way conversation powered by ElevenLabs Conversational AI and Twilio.
AI Everywhere — right-click context intelligence
Every number, chart and claim is right-click explainable. Select any text and choose 'Ask AI' — the assistant answers in the context of that exact data point. Contextual intelligence at the pixel level.
3D Digital Twin for client rehearsal
Rehearse the full conversation with a real-time 3D clone of the client before any sensitive meeting — animated, lipsync'd, and grounded in that client's actual DNA. Powered by 3DClone.me and MetaRoom.city.
'Why This?' — full decision traceability
Every recommendation traces to its API source. An animated decision tree shows exactly which market signal triggered the alert, which DNA value it conflicts with, and why the swap was chosen.
OPA Policy Engine — compliance by design
Every AI output passes through Open Policy Agent before reaching the RM. Hallucination detection, suitability checks, concentration limits, sanctions blocks, and citation enforcement — enforced in milliseconds.
Client DNA — the deepest personalization layer
Each client has a DNA profile built from CRM notes, behavioral patterns, family context, and stated values — extracted by LLM with Explicit / Pattern / Inferred confidence scores.
Deterministic alert engine — auditable by design
The core decision logic is pure deterministic TypeScript, not AI. Fully auditable. The LLM only writes language — the decision is always code.
Team
Built with and supported by
.svg){kind=logo}
