InflationIQ homepage — AI-powered inflation intelligence platform
Country + commodity forecast with confidence intervals
CPI trend visualization with risk scoring
Personalized investment signals — BUY/SELL/HEDGE/HOLD

InflationIQ — Inflation, Decoded 🌐📈

Inspiration

Inflation is one of the most powerful yet least understood forces shaping everyday financial decisions. Whether you're a founder planning expenses, an analyst hedging a portfolio, or simply someone trying to understand why groceries cost more — the data exists, but the intelligence layer connecting raw economic indicators to actionable insight does not.

I wanted to build something that bridges that gap: a tool that doesn't just show you inflation numbers, but tells you what they mean and what to do about them.

The name Orion — this hackathon's namesake — symbolizes navigation into the unknown. InflationIQ is exactly that: a navigation system for economic uncertainty.

What It Does

InflationIQ is an AI-powered inflation forecasting platform that lets you:

Pick a country + commodity (e.g. India + Natural Gas, US + Oil)
Set a forecast horizon (3, 6, 12, or 24 months)
Get a real AI-generated analysis including:
- Predicted CPI trajectory with confidence intervals
- Risk score and volatility regime classification
- A typed BUY / SELL / HEDGE / HOLD investment signal with tickers
- A full narrative explaining the economic reasoning

The free tier offers 3 analyses per day with no signup, making it accessible to anyone.

How I Built It

The stack is intentionally lean and fast to ship:

Layer	Technology
Frontend	React + TypeScript + TailwindCSS
Routing	TanStack Router
AI Engine	Claude 3.5 Sonnet (Anthropic)
Build Tool	Vite + Bun
Deployment	Vercel

The core intelligence loop works like this:

$$\text{Input}(\text{country}, \text{commodity}, \text{horizon}) \xrightarrow{\text{Claude 3.5}} {\hat{\pi}, \sigma, \text{signal}, \text{narrative}}$$

Where $\hat{\pi}$ is the predicted inflation rate, $\sigma$ is the volatility estimate, and the signal is one of the four typed action states.

The prompt is carefully engineered to keep all reasoning server-side, returning structured JSON with forecast, risk, signal, and narrative blocks — mirroring a real fintech API response format.

A live ticker across the top of the app displays real commodity and CPI data points (Gold, Wheat, Copper, DXY, US CPI, EU HICP, and more), giving users immediate market context before they run a prediction.

Challenges

1. Structured AI output reliability Getting Claude to consistently return well-formed JSON with all required fields across diverse country/commodity combinations required multiple prompt iterations and strict output validation.

2. Making forecasts feel credible without being misleading Economic forecasting is inherently uncertain. I spent significant time designing the confidence interval display and risk regime labels so users understand they're getting probabilistic intelligence, not guaranteed predictions.

3. Shipping a full product solo under time pressure From blank canvas to a deployed, multi-page app with a pricing page, API docs page, and working predictor — all within hackathon hours. Every architectural decision had to be fast and clean.

What I Learned

How to engineer prompts for structured, typed AI outputs at scale
Designing UX that makes complex financial data feel approachable
The importance of progressive disclosure — showing just enough information at each step without overwhelming the user
How real fintech products structure their API surface (forecast → risk → signal → narrative)