Anime Agent

Inspiration

The idea stemmed from the frustration anime fans face when trying to find their next favorite show. Static recommendation lists often miss the personal touch. Inspired by the potential of AI agents and a love for anime, we decided to create an interactive anime recommender that users can simply chat with — just like a friend who knows what you’d enjoy watching next.

What it does

Anime Agent is an AI-powered conversational agent that gives users personalized anime recommendations based on their favorite shows. Users just send a message like “I liked Naruto,” and the agent responds with 2–3 smart suggestions, powered by a generative AI model through the ASI Mini API.

How we built it

Built with Fetch.ai’s uAgents framework to deploy and run autonomous agents.

Used ASI Mini Client to connect to an AI model capable of generating meaningful recommendations from natural language prompts.

Defined message-handling protocols and async response logic to make the interaction smooth and responsive.

Deployed and tested the agent on the Fetch.ai agent network.

Challenges we ran into

Module limitations: Certain libraries like pandas and sklearn couldn't run in the sandboxed environment, requiring a pivot in the architecture.

Interface confusion: The chat/message button in the Agentverse interface was not always visible, making it hard to test interactions.

Debugging async flow: Managing async messaging between agents while catching API failures gracefully needed multiple iterations.

Time constraints: Building a production-ready conversational agent in limited time was a tough but rewarding task.

Accomplishments that we're proud of

Created a fully working anime recommendation agent that runs autonomously on the Fetch.ai network.

Integrated a third-party AI (ASI Mini) seamlessly into the agent’s logic.

Successfully deployed a public-facing agent that is alive, discoverable, and chat-capable.

Designed a clean, extensible architecture for future improvements like image-based input or voice support.

What we learned

The basics of decentralized AI agent development with uAgents.

How to use AI APIs to make agents more intelligent and conversational.

Importance of writing robust, minimal code to fit within the platform’s constraints.

Real-world application of prompt engineering for domain-specific tasks like anime recommendations.

What's next for Anime Agent

Add support for search by genre, emotion, or mood (e.g., “recommend me sad anime”).

Enable image or character-based recommendations via computer vision integration.

Improve recommendation accuracy with a fine-tuned model trained on anime databases.

Add memory to the agent so it can hold short-term user preferences.

Launch a web or mobile front-end for wider public interaction outside Agentverse.

Built With

• agentverse
• asi
• asyncio
• fastapi
• python
• uagent