AI Pet Finder

WenheLI — Sun, 26 Jan 2025 06:40:43 +0000

This is a submission for the Agent.ai Challenge: Full-Stack Agent (See Details)

Teammates:

Zixiong Feng
Emma Wang

What I Built

Finding the perfect pet can be a long and overwhelming process, and we wanted to make it easier. That’s why we created this AI-powered pet matching agent—to help you connect with your ideal companion faster and with less hassle.

We know from personal experience how time-consuming it can be to browse through endless online listings, gather all the details about different pets, and try to figure out which one might be the best fit. We all wished there was a quicker, more streamlined way to do it. That’s why we built this tool. With just a few simple text inputs like your pet preferences (breed, size, temperament) and your zip code, our agent does the work for you and generates an excel sheet with currently available pets that meet your needs, complete with all the info you need to make an informed decision. It’s designed to save you time and help increase your chances of finding the perfect match, so you can focus on the joy of welcoming your new pet into your life.

From the technical perspective, we built two APIs to support our needs:

An API that connects with pet-finder to help us retrieve the essential information for making decisions in the following steps
An API that takes all information generated by GPT and aggregate them together to an excel that can be downloaded by users in the UI.

The Pet Finder agent flow created with Agent.ai:

Below is the exact steps we took in the agent.ai platform

Demo

Gtihub Repo: https://github.com/WenheLI?tab=repositories

Agent URL: https://agent.ai/agent/tzjat6apxaw3jz3k

Video Demo:

Below is a screenshot of the excel generated by our agent:

We want the user to use such an excel to better track the pets and their connections.

Agent.ai Experience

The Agent.ai platform is an excellent tool for prototyping ideas and building AI agents with a low-code approach. I really appreciate the concept of using templates and sharing them, which allows users to build upon others' work and accelerate development. However, we encountered some challenges during our experience:

Limited User Input and Output Options:
- In our project, we needed users to interact with a map and use a table format to submit enriched information. Additionally, we wanted to generate an Excel file that could be directly downloaded from the UI.
- Implementing the Excel download functionality required a significant amount of effort to find a suitable workaround. It would be highly beneficial if the platform could natively support these features, especially the ability to handle map interactions, table-based inputs, and file downloads seamlessly.
Limited Flexibility with GPT Responses:
- We found it challenging to directly access and manipulate values from GPT responses within the platform. Modifying or enhancing the content required external tools or cumbersome workarounds like building an API to carefully take care of it.
- Having the ability to write small scripts or customize the payload within the platform would greatly enhance flexibility. This capability would allow developers to fine-tune responses and perform dynamic modifications without needing to step outside the platform.

Overall, while the platform offers a fantastic starting point for AI agent development, addressing these challenges would significantly improve the user experience and expand its potential applications.

At the end, also, a great thanks to my teammates Emma & Zixiong. This project can not get shaped like this without their helps

Reduce overhead for data serialization between Wasm & JS (Help!)

WenheLI — Wed, 03 Jun 2020 01:48:13 +0000

I am trying to export Rust gif library to js runtime by using wasm to get a high-performance gif encoding/decoding library (repo). However, when running benchmarking, I found huge overhead for data serialization in the process (15ms for decoding & 300ms+ for data serialization).

Any ideas on how to address it, folks?
Help and discussion are appreciated 👍

Forem: WenheLI

AI Pet Finder

What I Built

Demo

Agent.ai Experience

Reduce overhead for data serialization between Wasm & JS (Help!)