04463ntm a22004937i 4500 000727640 CZ-PrVSE 20260315164443.0 m d cr n|||||||||| 260315s2025 xr fsbm 000 0 eng d NEZPRACOVANÝ IMPORT ABA006 cze ABA006 ABA006 rda Víteček, David ISIS:38270 dis Next-Generation Wealth Advisory : A Multi-Agent Hybrid AI Model eng Next-Generation Wealth Advisory : A Multi-Agent Hybrid AI Model David Víteček 2025 ?? stran : digital, PDF soubor Vedoucí práce: Miroslav Umlauf Závěrečná práce - Institut celoživotního vzdělávání (MBA)—Vysoká škola ekonomická v Praze. Fakulta informatiky a statistiky, 2026 Obsahuje bibliografii Textový (vysokoškolská kvalifikační práce) Rok obhajoby 2026 The accelerating emergence of AI-native advisory platforms is reshaping the wealth-management industry, creating both an opportunity and a strategic necessity for institutions facing rising client expectations and a shrinking advisor workforce. By 2034, the U.S. market is projected to experience an advisor shortfall of nearly 30%, while investors increasingly demand real-time, hyperpersonalized, and digitally delivered guidance. This thesis responds to these pressures by designing, implementing, and validating a Hybrid AI Advisor—a multi-agent, modular system that enhances traditional advisory workflows with scalable AI capabilities while preserving transparency, auditability, and regulatory alignment. The system integrates several coordinated components: a GPT-based conversational agent that converts unstructured dialogue into structured risk profiles and financial objectives; a rule-based AI Portfolio Selector that maps these profiles to suitable investment strategies with deterministic, auditable logic; and a market-signal engine incorporating indicators such as Squeeze Momentum and Parabolic SAR to contextualize portfolio decisions. These inputs feed into a Python-based rebalancing engine that produces reproducible allocation updates and transaction-ready instructions. Quality and coherence are ensured through a dual validation layer combining cosine-similarity benchmarking and peer-LLM review. All modules operate within a multi-agent orchestration framework enabling transparent data flow, modular extensibility, and human-in-the-loop oversight. A fully functional prototype—implemented using GPT-based agents, MANUS orchestration, Python analytics, and a web-based client interface—was developed and published on GitHub as part of this research. Validation in a simulated enterprise environment demonstrated high consistency, explainability, and workflow traceability, confirming the technical feasibility of a “one-client–one-AI-advisor” operating model. The empirical findings show that hybrid architectures outperform traditional advisory processes in scalability, personalization, and error reduction, while remaining compatible with supervisory review and compliance requirements. The thesis concludes that hybrid multi-agent systems represent a credible and strategically advantageous direction for next-generation wealth management. As validation mechanisms mature and regulatory frameworks adapt, advisory workflows are expected to shift progressively toward AI-driven, self-validating models with human oversight reserved for exceptional or high-impact decisions. Způsob přístupu: Internet Hybrid AI Financial Advisor LLM Multi-Agent Systems Agent Orchestration Portfolio Optimization Validation Loop Umlauf, Miroslav ISIS:145945 ths Ngenda, Duncan opn Vysoká škola ekonomická v Praze. Fakulta informatiky a statistiky kn20010709399 dgg https://insis.vse.cz/zp/94721/podrobnosti VŠKP v InSIS https://insis.vse.cz/zp/94721 Hlavní práce https://insis.vse.cz/zp/94721/posudek/vedouci Hodnocení vedoucího https://insis.vse.cz/zp/94721/podrobnosti dc:identifier NEPOSILAT VSKP vse94721 260312 94721