JookieBot

• 

• 

• 

• 

• 

• 

Inspiration

JookieBot started as something deeply personal — a digital tribute to my childhood best friend, Jookie, who migrated from Jamaica when we were kids. He was my brother in spirit: always curious, always exploring, always pushing me to see the world differently. Years later, as I built my career in computing and finance, I realized I could bring that same energy back — not as a person, but as an intelligent assistant built with the warmth, wisdom, and curiosity that defined him. So I reinvented Jookie through code — a fusion of Caribbean soul, financial sense, and AI precision.

What It Does

JookieBot is an AI-powered financial and life assistant with a Jamaican twist. It helps users:

Track and analyze bills and spending with interactive visual dashboards.

Get personalized financial advice through AWS Bedrock agents.

Learn money management and budgeting habits in a relatable, culturally intelligent way.

Chat in real-time through the JookieBot persona — combining humor, empathy, and smart analysis.

Underneath the friendly exterior is a layered intelligence system that pulls data from Pinecone, Bedrock Knowledge Bases, and Lambda functions — creating a hybrid AI that feels human but thinks like a strategist.

How We Built It

JookieBot was built with a TypeScript + React frontend that powers a dynamic financial dashboard. The backend started on AWS Lambda + Bedrock, with Cohere embeddings and Titan models for reasoning. When AWS Bedrock services became unstable during the hackathon, I migrated all vector data and knowledge base logic to Pinecone for reliability and speed. The UI includes modules like:

Board.tsx and CenterDisplay.tsx — the main AI + analytics hub

DataHubModal.tsx — where users input and manage bills

DebtCalculator.tsx — for financial payoff simulations

AnalyticsDisplay.tsx — for visual trend analysis

and a full JookieBot AI chat layer integrated through AWS runtime APIs.

Challenges We Ran Into

This build wasn’t smooth sailing:

AWS Bedrock changed APIs twice during the hackathon, breaking key knowledge base links.

Lambda integrations were tough to deploy and required multiple IAM and role fixes.

AWS outage took down all my compute and knowledge bases mid-development — forcing an emergency migration to Pinecone.

Rebuilding embeddings, reindexing documents, and reconfiguring agent endpoints took hours under pressure — but every setback refined the system’s architecture.

Accomplishments That We're Proud Of

Successfully recreated the spirit of a real person through AI — blending emotional intelligence with technical design.

Built a fully modular financial analytics system with real-time bill tracking, debt simulation, and AI coaching.

Migrated live data from AWS to Pinecone during production — with zero data loss.

Designed a UI that feels like a game board for your finances, making money management approachable and visual.

What We Learned

Always architect for resilience and portability — cloud APIs can change without warning.

AI is most powerful when it’s personal — grounding complex systems in culture, story, and emotion makes people connect to it.

Bedrock and Pinecone make a great hybrid: Bedrock for reasoning, Pinecone for embeddings and retrieval speed.

Building an AI with personality is about more than chat — it’s about experience design, emotion, and memory.

What’s Next for JookieBot

Next steps include:

Deploying JookieBot Agents across finance, travel, and education verticals.

Expanding the JookieBot Knowledge Graph across multiple languages and cultural contexts.

Integrating voice and video personas so users can talk to Jookie like a real friend.

Launching JookieBot Cloud, a Bedrock + Pinecone hybrid backend for developers who want to create culturally aware AI assistants.

Even through the chaos — broken APIs, outages, and all — JookieBot proved one thing: Friendship and innovation share the same spirit — they both adapt, grow, and keep showing up when you need them most.

Built With

• amazon-web-services
• and
• api-gateway
• aws-bedrock
• aws-kms
• aws-lambda
• cloudflare
• cloudwatch
• cohere
• framer-motion
• github
• iam
• json-+-localstorage
• node.js
• pinecone
• pinecone-vector-db
• proxmox
• react
• recharts
• secrets-manager
• tailwindcss
• terraform
• titan-embeddings
• typescript-(vite)
• vs-code