InvestSuite — Hackathon Project

An adaptive investing interface where the UI reshapes itself based on who you are and what's happening right now. Built for the Next Level Challenge.

Today's apps are databases with buttons. They ask every investor to adapt to the software. What if the software adapted to the investor?

Same person, different moment — different interface.

---

Architecture

┌─────────────┐    context signals    ┌──────────────┐     prompt     ┌───────────┐
│  Investor   │ ────────────────────► │    Brain     │ ─────────────► │  Claude   │
│  opens app  │                       │  (ASP.NET)   │ ◄───────────── │   API     │
└─────────────┘                       │              │  JSON response └───────────┘
                                      │  Assembles:  │
                                      │  - narrative │
                                      │  - tone      │
                                      │  - density   │
                                      │  - widgets[] │
                                      │  - chips[]   │
                                      └──────┬───────┘
                                             │
                                             ▼
                                      ┌──────────────┐
                                      │    Face      │
                                      │  (Next.js)   │
                                      │              │
                                      │  Renders     │
                                      │  whatever    │
                                      │  the brain   │
                                      │  sends.      │
                                      └──────────────┘

Tech Stack

Layer	Technology
Frontend	Next.js 14 · React 19 · Tailwind CSS 4 · Radix UI · Framer Motion
Backend	ASP.NET Core 9 · System.Text.Json · SQLite
AI	Claude API (Anthropic)
CI	GitHub Actions

---

Local Setup

1. Backend

cd backend/src/InvestSuite.Api
cp appsettings.example.json appsettings.json

Edit appsettings.json and add your Anthropic API key:

{
  "Claude": {
    "ApiKey": "sk-ant-api03-YOUR_KEY_HERE",
    "Model": "claude-haiku-4-5-20251001"
  }
}

dotnet run

2. Frontend

cd frontend
cp .env.example .env.local

Set required variables in .env.local:

AUTH_SECRET=<generate with: openssl rand -base64 32>
NEXT_PUBLIC_API_URL=http://localhost:5062

npm install
npm run dev

Open http://localhost:3000 — sign up with email/password or Google.

---

How the Brain Works

Given context signals (market volatility, session frequency, time of day, emotional state), the brain picks:

Decision	What it controls
Tone	Color palette, typography, spacing
Density	Layout grid, number of widgets, card sizing
Widgets	Which components appear, in what order
Chips	3 contextual next-step suggestions
Narrative	1–3 sentences — the first thing the investor reads

7 Widgets

Category	Widget	Description
Information	portfolio_overview	Full portfolio summary, allocation chart
Information	stock_card	Single holding detail (simple/standard/full)
Information	explanation_card	Education, reassurance, catch-up, insight
Visualization	historical_chart	Price/performance over time
Visualization	volatility_gauge	Market temperature (low/medium/high)
Visualization	comparison_view	Side-by-side two holdings
Action	order_confirmation	Buy/sell confirmation (only on explicit request)

19 Contextual Chips

The brain picks 3 per response from a fixed pool:

Category	Chips
Portfolio	How am I doing? · Show my allocation · Show my returns
Research	Tell me about {holding} · Compare my ETFs
Education	What are ETFs? · What are dividends? · What is P/E ratio? · How does diversification work? · Stock vs ETF?
Market	What happened? · Am I okay long-term? · Show the bigger picture
Catch-up	Catch me up · What did I miss? · Show my dividends · Anything new?
Navigation	Back to overview · I still want to sell

---

Demo Scenarios

Scenario	Investor	What happens
Calm Tuesday	Beginner	Low volatility, evening check-in. Simple overview, educational tone.
Crash Morning	Beginner	-6.2% week, 7:42am, 2nd session. Reassuring tone, perspective widgets.
Returning	Expert	94 days absent, +4.2% growth, dividends. Welcoming catch-up story.
Research Mode	Expert	3rd session today, searching semiconductors. Dense, focused layout.

---

API

Dashboard (proactive)

GET /api/brain/dashboard/{investorId}?scenario={key}

Interaction (reactive)

POST /api/brain/interact/{investorId}
{ "message": "...", "scenario": "...", "conversationHistory": [...] }

Both return the same response shape. The frontend doesn't distinguish.

---

Case Rules (Non-Negotiable)

1. No investment advice — inform, don't recommend
2. No gamification — no points, badges, streaks
3. Investor stays in control — no manipulation or urgency
4. Beyond the chat bubble — the UI itself must change shape

---

Team

• Alken Rrokaj
• mesyrob
• Viktor Gakis
• Thomas Vrolix