Personalized investment consultancy demands continuous market analysis, risk management, and transparent communication. This paper introduces PeterAI, a multi-agent framework leveraging Large Language Models (LLMs), chain-of-thought (CoT) reasoning, and portfolio optimization to deliver personalized financial advice via WhatsApp. The system integrates specialized agents (fixed income, equities, REITs, multimarket funds), supports multimodal inputs (text, audio, images), and processes real-time financial data. In a seven-month study (June–December 2024) with 120 clients, PeterAI-managed portfolios outperformed human advisors by achieving significantly higher returns ( \(8.29\,\%\ \pm \ 0.30\,\%\) vs. \(2.57\,\%\ \pm \ 0.99\,\%\) ), reduced volatility ( \(1.45\,\%\ \pm \ 0.32\,\%\) vs. \(2.64\,\%\ \pm \ 0.53\,\%\) ), and superior Sharpe ratios ( \(2.67 \pm 0.85\) vs. \(-0.85 \pm 0.36\) ). Key contributions include: (i) a MAS-LLM architecture for scalable personalization; (ii) an error-handling pipeline with ambiguity resolution; (iii) auditable explainability via CoT. Results demonstrate PeterAI’s potential to transform financial advisory through AI-driven automation and regulatory-compliant transparency.