Fetch - Ai Hackathon - Application - IIT-M
Fetch - Ai Hackathon - Application - IIT-M
Fetch.ai
Agenda
1. Introduction
2. uAgents, AI Engine, Agentverse, DeltaV
3. Walkthrough
Let’s talk about AI, today.
AI -> AGI -> ASI
LLMs lack ‘action’
● LLMs can’t act
● Example: ChatGPT
● LLMs need to act.
● To act = Agentia (latin)
What are AI Agents?
● Pieces of code that perceive their environment and take actions in
order to achieve specific goals
● Usually, interact autonomously with other agents
What is Fetch.ai, and where does it fit in?
Components
1. AI Agents are independent decision-makers that connect to the
network and represent data, APIs, services and people
2. Agentverse serves as a development platform for these agents
3. AI Engine links human input to AI actions
4. DeltaV is a simple user interface for non-technical users
5. Fetch Network underpins the entire system, ensuring smooth
operation and integration
Integrations, Solutions, Applications
● Integration: Third party tools/apps/APIs that can be directly
integrated with uAgents + DeltaV. Example: Google Docs, RapidAPI,
etc.
Integrations, Solutions, Applications
● Integration: Third party tools/apps/APIs that can be directly
integrated with uAgents + DeltaV. Example: Google Docs, RapidAPI,
etc.
● Solution: Combination that ‘chains’ multiple integrations to achieve a
use-case on DeltaV. Example: Fetching latest data from a particular
API from RapidAPI and feeding it into Google Docs.
Integrations, Solutions, Applications
● Integration: Third party tools/apps/APIs that can be directly
integrated with uAgents + DeltaV. Example: Google Docs, RapidAPI,
etc.
● Solution: Combination that ‘chains’ multiple integrations to achieve a
use-case on DeltaV. Example: Fetching latest data from a particular
API from RapidAPI and feeding it into Google Docs.
● Application: An end-to-end application that uses uAgents. Doesn’t use
the DeltaV interface. Is a standalone app (example: Blockagent)
Quick Walkthrough: From Zero to DeltaV
Step One: uAgents
You can stop following the guide when you reach the section ‘Booking a
table at a restaurant’.
Step Two: Agentverse
1. Create an Agentverse account on https://agentverse.ai
2. Create an Agent on online IDE of Agentverse (Link to guide)
3. Use the mailbox service to create a local agent (Link to guide)
You need to use mailbox service if you are using a python module apart
from the ones listed here. It is IMPORTANT to note that due to security
reasons, Agentverse doesn’t allow import of all modules. But you can run
your agent locally and connect it to Agentverse via Mailbox.
Step Three: DeltaV
1. Register a Hugging face API agent as a service (Link)
2. Now try your API/Problem Statement and connect it to DeltaV.
For more examples, you can check the integrations repository. Remember
that you DO NOT need to build your own UI. Your project must be
accessible from DeltaV.
OR
Examples
If you’re using your custom frontend
● You don’t need to raise a PR
● You just need to submit the link of your project (you can host it on free
cloud services) in a Google Form (to be shared later)
● You will need to demo the project at Paradox (date TBD)
If you’re using Agentverse/DeltaV
● You need to raise a PR
● You just need to submit the link of your repo in a Google Form (to be
shared later)
● You will need to demo the project at Paradox (date TBD)
Here’s what we would be judging you on
● Functionality: Is your application functional end-to-end?
● Creativity: Are you merely ‘integrating’ something - or are you
presenting a business use-case as well?
● Code Quality: Is your code clean, organized, and well-commented?
Directory Structure
README file
● Project name
● Description of the project
● Instructions to run the project
● Use-case example
● Special considerations, if any
Yes, because there are security restrictions. You can create a local agent
and connect it to a mailbox
> Can I use any additional API/integration and connect two agents
Nope.