The Dawn of Microagents
The increasing sophistication of AI models like ChatGPT is a symptom of the dangerous levels of data centralization we have reached in Web2. The more data is collected and controlled by the Big Five, the more highly optimized machine and deep learning models they create, and the less individual privacy we will all have. Fetch.ai’s raison d’être has always been to disrupt that status quo by providing a decentralized, peer-to-peer alternative with our Autonomous Economic Agents (AEAs) framework. Now, we are getting one step closer to this goal with the release of Microagents.
Benefits and Challenges of AEAs
Fetch.ai’s AEAs are software programs capable of acting independently and autonomously in the pursuit of their owner’s economic interests. They have the agency to get information from their environment, to communicate and cooperate with each other, and to do all this without direct instructions from the user.
The Autonomous Economic Framework represents in fact the communication protocols that facilitate the creation of proactive agents that search and discover, negotiate and transact. It makes it possible for multi-agent systems to be established, which in turn can solve many tasks that are too difficult or even impossible for either individual agents or monolithic structures.
Allow us to give you an example. Imagine a weather station deploys a weather data agent that sources information from its environment and looks for the best price to sell it. On the other hand, a buyer of weather data — be it a weather forecast app or a self-driving car, in turn, creates a weather client agent. The agents do the entire process of searching the data, looking for a supplier / a client, negotiating the price, and performing the transaction.
In other words, if you want to easily introduce automation into processes — optimize human resources or remove centralized intermediaries, and thus revolutionize a plethora of industries, AEAs are the tool to use. What is more, Fetch.ai’s agents will allow you to achieve all this by simultaneously ensuring that you and your users retain full ownership of your data and privacy.
While it’s still early days for Fetch.ai’s AEAs, we recognize that their multi-layered structure may prove challenging. Agents are composed of multiple core components that include skills, connections, contracts, and protocols, and require these resources to be implemented in a hierarchy. Although they were designed to offer high levels of modularity and code reusability, they have proven to be quite challenging for developers.
Thus, we’re taking steps to reduce the steep learning curve and lower the entry barrier for new developers to join our AEAs ecosystem by introducing Microagents.
Why do we need Microagents ?
Fetch Microagents will be for Web3 what microservices are for Web2.
Not quite understanding what that means? Let’s break that statement into pieces.
Imagine an application where all the processes within are part of one single block of code and run as one single service. That is what is called a monolithic architecture. It requires all processes to be developed, deployed and updated together, and to have the same dependencies to external packages. Consequently, if one process of the application experiences a spike in demand, the entire architecture must be scaled. In time and with the evolution of computer science, this approach turned out to be quite problematic, especially given the growth in size and complexity that most applications have experienced.
To address this, software developers came to the conclusion that such a design can bring more issues than benefits, and came up with the microservice architecture. Here, an application is like a collection of smaller applications. Each process represents an independent component that delivers a service and only executes one function. Also, each separate building block is independently deployed, run, and updated. That way, the development and maintenance of complex applications become much easier and quicker.
As you have already guessed, the microservice architecture offers a playground for much more experimentation, testing new ideas, and innovation than the monolithic one could ever have. And this is what we intend to achieve thanks to Microagents.
What are Microagents?
Microagents represent a very limited subset of the capabilities of the Autonomous Economic Agents. They are purposefully limited in order to help developers quickly reach the minimum viable product stage and add complexity at a later stage where required. In this context, Microagents will act as a gateway to the broader AEA framework, which can ultimately serve more general use cases.
Currently, Microagents are a separate protocol from the AEAs, but we will be working toward enabling the communication between the two. Implemented as a light Python library, Microagents facilitate the creation of modular and decentralized applications by smoothing the path of testing and prototyping. They are much easier to get up and running than AEAs, and provide native horizontal scalability.
Moreover, we are releasing Microagents together with a dedicated communication protocol that codifies how they will register and talk to each other. Similarly to the Domain Name System (DNS) acting as the phone book of the Internet, “the agent almanac” will act for Microagents. And to make this whole announcement even more fascinating, the Microagents are also coming with easy-to-implement and reusable app templates.
Our world needs more open-source privacy-preserving tools in order to give democracy and healthy competition of ideas a fair chance. Fetch.ai has been consistently delivering such instruments and now goes one step further with the Microagents. These peer-to-peer microservices, made for Web3, are simple, composable, and scalable. Their main goal is to boost developer productivity and accelerate time-to-market for new ventures and features. From automating mundane and repetitive processes to enabling complex and sophisticated decision-making, the power to leverage microagents is entirely in your hands.
For detailed information on the inner workings of micro agents head over to our documentation https://docs.fetch.ai/uAgents/