Gift Guru

Inspiration

Recently, I found myself selecting gifts for friends and coworkers, and I began contemplating the potential for developing an AI consultant to guide users through this process.

What it does

The AI consultant acts as a coach, posing questions about the recipient, your relationships, and intentions. Based on your responses, it offers gift recommendations and suggests ways to personalize the present.

How I built it

I started by researching various resources, including articles and existing LLM models, to formulate a process for brainstorming gift ideas and making selections. After analyzing several articles, I identified key areas to collect information, such as the interests of the recipients, their passion, your relationship with the recipient etc.

Challenges I ran into

Integrating web search for recommended gifts into the PartyRock environment proved challenging. To overcome this, I plan to move the prototype to the AWS environment and integrate it with a web search and recommendation engine API.

Another challenge I faced was ensuring that the application maintains consistency by asking the same set of questions each time. Simultaneously, I aimed to accommodate variations in how the questions were formulated, all while keeping the flow of the conversation flexible.

Accomplishments that I am proud of

I take pride in creating a prompt that consistently asks a predetermined set of questions while remaining flexible in phrasing and adapting to user input. The system provides open-ended recommendations based on the context provided by the user.

I applied this application in a real-life scenario and it assisted me in selecting a gift for a friend of my 3-year-old daughter!

What I learned

I learned how to design a prompt that efficiently collects required information from a user in a predictable manner, while still allowing flexibility in conversation and recommendations.

What's next for Gift Guru

The upcoming steps include deploying the app in the production environment on AWS and integrating it with Bedrock. I plan to build a scalable serverless chat system capable of handling thousands or millions of users and integrate it with a search engine API to provide links to the recommended gifts. This setup will enable monetization through offering affiliated links to online stores, like Amazon.

Built With

• partyrock