Find a Helper

The smart task platform with a built-in AI advisor. Our chatbot helps you post, find, and strategize tasks, ensuring every job is handled with expert-level guidance and efficiency.

Comment

Inspiration

Picture this: You just moved to San Francisco from New York. You have no friends or acquaintances nearby. Your boxes are piled high, you need a quick helper, and you know no one! How about neighbors? Potentially, but you know people are busy; there is still a high chance you are by yourself. On the flip side, imagine you are a college student short on money, looking for a "quick buck." You want to walk dogs or tutor, but you don't know where to find those jobs safely or easily. I realized that while many platforms exist, they often feel impersonal or leave users guessing how to describe what they need. I built Find a Helper to bridge this gap. I wanted a platform that doesn't just list jobs but actually guides you through the process. By integrating an AI chatbot that helps you post needs, find opportunities, and offer advice on tasks, I’m turning stressful solo moments into supported experiences.

What it does

Find a Helper is a community-driven web application where help is just a conversation away. It is a platform for users to post tasks they need help with and for helpers to find tasks they can do with AI integrated to help them with their tasks.

  • The AI Advisor: A built-in chatbot that helps users frame their tasks clearly and gives helpers advice on how to best complete the job.
  • Task Marketplace: A streamlined interface for posting chores, errands, or professional help.
  • Student/Gig Portal: A dedicated space for those looking to earn money by helping their neighbors.
  • Conversational Task Creation: Instead of filling out long forms, users simply tell the AI, "I need help moving a couch this Saturday," and the AI drafts the title, description, and reward.
  • Smart Task Recommendations: The AI analyzes a helper's bio and expertise (e.g., "Gardening" or "Moving") to provide a ranked list of recommended jobs with explanations of why they are a good fit.
  • Visual Match Scoring: The platform uses a color-coded system to show compatibility: Purple (High Match), Orange (Medium), and Red (Low).
  • AI Insights: Users can ask the AI for task statistics or price suggestions to ensure fair market rates for every job.

How I built it

The project is built using a Python-based backend designed for rapid deployment and intelligent interaction:

  • Backend: Developed with Python and Flask, and handle user and task data via SQLite.
  • AI Integration: Powered by the OpenAI API, the assistant is equipped with custom tools like get_recommended_tasks to bridge the gap between static data and personalized user experiences.
  • Frontend Logic: Implemented dynamic UI elements, such as the match-score color coding, to provide immediate visual feedback to helpers.

Challenges I ran into

Integrating the AI was the most difficult part of the build because I wanted a proactive agent, not just a chatbot.

  • Connecting Data to AI: I had to bridge the gap between natural language and my SQL database. I used an MCP-inspired approach to standardize how the AI "discovers" and uses tools like search_tasks and get_tasks_statistic.
  • Mapping User Intent: A major hurdle was ensuring the AI correctly translated vague requests (e.g., "Find me work") into structured queries using my get_recommended_tasks function.
  • Preventing Hallucinations: I had to implement strict error handling to ensure that if the AI suggested an invalid price or category, the backend would catch it and prompt for clarification rather than crashing the task creation flow.

Accomplishments that I'm proud of

  • Building a Functional AI Agent: I successfully moved beyond a simple "chat-wrapper" by implementing an MCP-inspired architecture. This allows the AI to act as a true agent that can query my database, calculate statistics, and offer real-time advice.
  • Solving the "Two-Sided" User Experience: I created a seamless flow for two very different users, the "Newcomer" who needs help and the "Student" who needs work, ensuring the AI supports both their unique needs through the same interface.
  • Intelligent Matching System: I am particularly proud of the Smart Match Score. Developing a tool that uses AI to rank jobs by matching a helper’s specific bio to task requirements, and then visualizing that with color-coded UI, significantly improves the efficiency of the marketplace.
  • Conversational Task Creation: I successfully automated the friction of "form-filling." My implementation allows a user to post a task through natural conversation, where the AI intelligently identifies missing details like rewards or dates before finalizing the post.
  • Robust Backend Integration: I built a backend that handles automated database initialization and secure tool-calling, ensuring that the AI has direct, safe access to local context without requiring manual database management.

What I learned

This project deepened my understanding of building tool-augmented AI agents. I learned how to connect an LLM to a local database so it can perform real-world actions like searching tasks and generating statistics. It also highlighted the importance of UX in AI, showing the user why a task was recommended is just as important as the recommendation itself.

What's next for Find a Helper

  • AI Profile Optimizer: Helping users write better bios to get more matches.
  • Safety & Moderation: Using AI to flag inappropriate or "scam" task postings.
  • Multilingual Support: Breaking language barriers to help non-native speakers find work and help.
  • Forum for the community
  • AND MORE
Share this project:

Updates