Hackathon Showcase

EQUITYA

We launch Finance AI agents at scale. Starting with an full stack AI wealth agency.

3 members

Project Description

Our project delivers an end-to-end AI Wealth assistant that replaces human-heavy, fee-heavy wealth management with a suite of autonomous agents. We demonstrate full Execution & Functionality: users connect accounts via Powens integration, view a live wealth dashboard, and interact with a Wealth AI Agent that routes queries to specialized sub-agents (tax, investment strategy, cashflow, risk). Each returns structured analysis and actionable recommendations, showing clear AI & Agents Usage well beyond a single LLM call.

We address a large, well-defined problem: traditional advisors charge ~1% AUM and prioritize the already-wealthy, while people with €100k–€5M are underserved. Our approach delivers private-bank intelligence instantly and at a fraction of the cost, demonstrating strong Problem Clarity & Market Impact.

To prove Scalability, we built two company-level agents:
– an Outbound Agent that detects prospects using public signals (IPO filings, funding rounds, job changes),
– an Inbound Content Agent that automatically publishes educational wealth content.
Together, these automate core growth functions, enabling a credible one-person firm model.

Our Demo Narrative is coherent and complete: problem → product → specialized agents → growth engine → vision.

Partner Tech Usage is central: Powens (bank data aggregation), ElevenLabs (voice discovery agent), Qdrant (Knowledge DB agent), Mistral & Gemini (reasoning LLMs), plus Figma, Lovable and Fal for product/UX/UI.

Technologies Used

AI & Agents: Mistral, Gemini, custom agent orchestration, Qdrant (RAG).
Integrations: Powens (financial data), ElevenLabs (voice).
Frontend: React/Next.js, TypeScript, Tailwind.
Backend: Node.js
Design: Figma (UI), Lovable (landing), Fal (Video & Image)
Hosting: Vercel

Agent Logic, Failure Modes & Next Steps

Wealth Agent Logic: Orchestrator agent classifies intent and delegates to tax, investment, cashflow, and risk agents; each uses Powens data + Qdrant knowledge + LLM reasoning.

Failure Modes: Missing data, API outages, and hallucinations handled via guardrails, fallback caching, uncertainty statements, and rule-based validation.

Next Steps:
In production mode with powens.
Improve inbound and outbound agent for acquisition growth.
Robust tax jurisdictions, advanced risk engine, compliance/suitability layer.
Integration of finance partners (PEA, PER, Life Insurance, CTO, SCPI, and other financial products…)
Build an agent able to execute financial orders for the client
Build an agent able to execute tax for the client.