CV_IntelliPA

Your AI Assistant for Tasks, Priorities & Reminders.

Comment

Inspiration

Managing tasks is not difficult; knowing what to focus on, when to act, and what can be safely postponed is.

We wanted to build a Virtual Personal Assistant that goes beyond being a reactive chatbot. The idea behind CV_IntelliPA was to create an assistant that understands tasks through natural conversation, remembers commitments, evaluates their urgency and impact, and proactively reminds the user before important actions are missed.

We chose WhatsApp as the primary interface because it allows users to interact with their assistant naturally without learning another productivity application.

What it does

CV_IntelliPA is an AI-powered WhatsApp Virtual Personal Assistant built using Amazon Bedrock and AWS services.

Users can simply message the assistant:

"Remind me to follow up with the customer tomorrow."

"I need to finish the AWS proposal by Friday."

"What should I focus on today?"

CV_IntelliPA can:

  • Create, update, complete, and reschedule tasks
  • Understand natural-language task descriptions
  • Analyze urgency and impact
  • Prioritize tasks based on context and deadlines
  • Schedule one-time and recurring reminders
  • Proactively send WhatsApp reminders
  • Track overdue tasks
  • Provide daily focus recommendations
  • Provide productivity and workload insights
  • Maintain task and conversation context

The key difference from a traditional chatbot is that CV_IntelliPA can proactively act based on time and task state, rather than waiting for the user to send another message.

How we built it

CV_IntelliPA extends an AWS open-source WhatsApp + Amazon Bedrock assistant architecture with a dedicated task-management and proactive-assistant layer.

The architecture uses:

  • WhatsApp / Meta Cloud API — communication interface
  • AWS End User Messaging Social - Receives WhatsApp messages and sends replies, interactive buttons, templates and login codes
  • SNS - Delivers inbound WhatsApp events to Lambda
  • Amazon API Gateway — webhook entry point
  • AWS Lambda + Python — application logic and orchestration
  • Amazon Bedrock — natural-language understanding, reasoning, urgency, impact, and priority analysis
  • AWS CDK (Python) / CloudFormation - Whole stack as code
  • Amazon DynamoDB — tasks, reminders, conversations, and user context
  • Amazon EventBridge Scheduler — one-time and recurring reminder scheduling
  • Amazon SES — Task/plan/meeting/weekly summaries and personal insights
  • AWS Bedrock Agent Core — Long-term per-user facts and preferences
  • Amazon CloudWatch — monitoring and logging
  • Lambda Function URL - Serves the IntelliPA PWA and its JSON API

We deliberately separated AI reasoning from deterministic business logic.

Bedrock interprets the user's intent and provides structured recommendations. The Python application validates those recommendations and performs the actual task management, database updates, and scheduling.

This gives CV_IntelliPA both the flexibility of generative AI and the reliability of traditional application logic.

Challenges we ran into

The biggest challenge was moving from a reactive chatbot to a proactive personal assistant.

A chatbot naturally responds when a user sends a message. A personal assistant needs to operate even when the user is not interacting with it.

We therefore had to design a separate scheduling and reminder mechanism that could:

  • Track future tasks
  • Trigger reminders at the right time
  • Handle recurring tasks
  • Detect overdue tasks
  • Support snoozing and rescheduling
  • Send proactive WhatsApp messages

Another challenge was determining how much responsibility should be given to the AI.

We did not want the LLM to directly control application state. Instead, Bedrock provides structured intelligence while Python validates and executes the resulting actions.

Natural-language ambiguity was another challenge. A request such as:

"Remind me to call Ravi sometime tomorrow."

requires the assistant to understand the task, determine the date, and decide whether it needs to ask the user for a specific time.

Accomplishments that we're proud of

We are proud of transforming a conventional conversational AI experience into a proactive task-oriented personal assistant.

Some of the key accomplishments are:

  • Built a WhatsApp-based AI personal assistant using Amazon Bedrock
  • Added persistent task management to a conversational interface
  • Implemented AI-driven urgency and impact analysis
  • Added proactive reminder scheduling
  • Created a serverless AWS architecture using Python and managed AWS services
  • Separated AI reasoning from application execution
  • Designed the foundation for contextual personal insights and workload analysis

Most importantly, CV_IntelliPA demonstrates how generative AI can be combined with traditional event-driven architecture to create an assistant that can remember, reason, schedule, and proactively communicate.

What we learned

We learned that building a useful AI assistant is about much more than integrating an LLM.

The LLM provides the intelligence to understand natural language and reason about tasks, but persistent memory, scheduling, event processing, validation, and application logic are equally important.

We also learned the importance of giving AI the right level of control. Letting the model recommend an action while allowing deterministic application logic to execute and validate that action provides a better balance between flexibility and reliability.

Most importantly, we learned that the difference between an AI chatbot and an AI assistant is proactivity.

What's next for CV_IntelliPA

We want CV_IntelliPA to evolve from a task reminder into a more capable personal productivity and decision-support assistant.

Planned capabilities include:

  • Calendar and email integration
  • Intelligent task dependency detection
  • Automatic follow-up identification
  • Daily AI-generated briefings
  • Weekly productivity insights
  • Context-aware task recommendations
  • Personal working-pattern analysis
  • Voice-based task creation
  • Integration with business applications
  • Smarter prediction of tasks that may become overdue

Our vision for CV_IntelliPA is simple:

Don't just remind me what I need to do. Help me understand what matters, when it matters, and what I should do next.

Built With

Share this project:

Updates

Submission history