Hoot: AI Stock Coach

Stock chart
Querying a section
Why a change happened in that section
Owl chatbot
Quiz page
Profile - Novice level
Analyst Level

Inspiration

Most people can read a chart. Most people can read a headline. Very few can connect them into a clear story for the exact slice of time they care about. We wanted that connection to feel like learning from someone patient and a little playful — so we built around an owl: wise, approachable, consistent. The chatbot lets learners ask the embarrassing follow-up questions in plain language instead of hunting through menus.

What it does

Pick a ticker, choose a horizon, drag a price window — get an AI explanation grounded in real headlines, in beginner or analyst tone.
Hoot anchors everything on the learning side: structured explain cards, the Market Game (driver identification, quick quiz, paper-trade framing), and free-form chat for "what does that mean?" or "why does this matter for my portfolio?" One voice, three surfaces.
The portfolio news digest merges RSS per holding into ranked story cards — so your news is connected to your actual book, not just the market at large.

How we built it

Next.js UI with an interactive Recharts workflow (select range → /api/explain).
Structured LLM JSON for explain + learn payloads; Groq / Gemini behind a small queue to respect rate limits.
Owl persona as the default teaching voice across UI copy and prompts (consistent tone: clear, curious, not preachy).
Chatbot layer for follow-up Q&A (context from the current symbol, selection, or lesson)—implemented as chat completions against the same model stack with guardrails.
RSS aggregation + caching for portfolio news; localStorage for flashcards and quiz history.

Challenges we ran into

Keeping token budgets sane when the chatbot, chart explain, and news digest all share one provider tier.
Making the owl feel like one teacher across structured cards and free-form chat (same vocabulary, same honesty about uncertainty).
JSON truncation when packing explain + learn + long chat turns—required tighter prompts and output caps per route.

Accomplishments that we're proud of

A recognizable mascot-led brand: the owl isn’t decoration—it’s the default teacher for the product story.
Explain → practice → chat in one loop: structured lessons plus conversational depth when something doesn’t click.
Shipping something that still runs on hackathon-friendly APIs by treating limits as part of the design.

What we learned

Persona + structure beat generic “assistant” vibes for education products.
Chat is where beginners actually ask the embarrassing questions—worth budgeting tokens for.
Reliability (parse success, timeouts) is as important as clever prose for demos.

What's next for Hoot

Richer chat memory per ticker/session (still privacy-conscious).
Owl-guided tours (first-time user paths) and suggested questions after each explain.
Stronger evaluation: which explanations and chat turns actually reduce confusion.