← Todos os casesCS—003 · DOSSIÊ DE RESULTADO

Project Alpinist: Multi-Agent AI for Investment Management

Case No.
CS—003
Sector
Finance & Banking
Scope
Conversational Agent · Automation
Stack
Tess AI · Multi-Agent Systems · N8N · Supabase SQL · Comdinheiro API
-60%
asset management time
24/7
analysis availability
01 / The Challenge

WealthPeak investors faced high technical complexity when managing their portfolios, relying on spreadsheets and dense menus. The lack of an immediate correlation between market news and the impact on personal assets caused insecurity and slow decision-making. In addition, executing portfolio adjustments was a manual process that drove away less experienced investors due to technical jargon.

”— NOTA DE CAMPO
“The complexity of financial data prevented agile, clear, and intuitive investment management.”
02 / The Solution

We developed Project Alpinista, a Multi-Agent AI ecosystem natively integrated into the platform's chat. An Orchestrator Agent triages user intentions, directing them to an Analyst Agent — which cross-references Comdinheiro data with the portfolio in real time — or to a Trader Agent, which executes operations via SQL on Supabase. The solution translates complex data into accessible language, enabling complete asset management via a conversational interface, without leaving the application.

S.1
Orchestration Architecture
Central engine for intent identification and intelligent routing between specialist agents.
S.2
Financial Data Integration
API connection with Comdinheiro for real-time ticker capture, news, and sentiment analysis.
S.3
Operations Automation (Trader)
Secure SQL functions in Supabase to enable adding and removing assets via voice or text commands.
S.4
Natural Language Refinement
Fine-tuning language models to ensure educational communication, eliminating technical jargon for the end user.