2020 has been a year of vast learning. We launched our incentivized testnet program which gave us great insight into the areas we can improve and develop ahead on the mainnet v2.0 launch. We now want to put these learnings into practice by developing the ledger and building upon it through improving our collective learning and agent frameworks.
By the end of Q1, we want to achieve a fully integrated system between our frameworks and the ledger. This greatly enhances the FET token utility, and enables anyone to participate in the network operations in many ways: operating validators, taking part in governance decisions, running search and discovery network components and building, operating and distributing agents and agent based applications. It’ll also be the first time that all the parts of the network are fully functional together, and we’re super excited!
In this article, we look at the key parts of mainnet v2 and cover just some of the things planned for the next few months. This includes our amazing agent framework, where you can build autonomous digital representatives of just about anything in fields as diverse as mobility, climate management, energy or healthcare, as well as the underlying ledger, search-and-discovery and powerful collective learning framework:
The Fetch.ai decentralized machine learning platforms, including all our frameworks, are built upon the same thing — our ledger. Designed as the perfect home for our technologies and our agents, it enables secure sharing, connection and transactions based on any data globally. It forms the bedrock of the greater Fetch.ai network so we are always looking for ways to improve it to make the network stronger, more secure and more accessible for all the amazing things being built, now or in the future. A few of the ways we are doing this in Q1 are:
- Mainnet v2.0: mainnet is scheduled to be released in March 2021
- Includes random beacon integrated with Tendermint for faster and more secure consensus.
- Beacon World testnet: an incentivized test net where FET holders create and operate nodes, become validators and perform other activities as we do mainnet launch rehearsals. The first part of this, Beacon World 1 has been launched. You can find more details here
- Secure pseudo-random number generator that is interoperable across chains. A huge amount of agent-based applications require such random numbers
As always, our work is open-source, and available for everyone to contribute to. We actively encourage this, and are improving our community participation incentives as we head to mainnet v2 and beyond.
Our agent framework provides the tools for creating autonomous economic agents. It is an exciting ongoing development aimed at making the creation of agents easier and easier, allowing more of you to use the Fetch.ai technology in a meaningful way. Some of the ways we plan to do this in Q1 include:
- Agent Framework V1 launch: full feature-complete, refined version 1 of the agent framework for commercial and community deployment on mainnet v2
- AEA GUI: release of local no-code interface to interact with Autonomous Economic Agents (AEAs)
- AEA CLI: full integration of IPFS, secure private key management
- AEA Registry: upgrades to expose additional information like AEA version compatibility and package specifications.
- ACN (Agent Communication Network): further reliability and security improvements to ready the ACN for continuous production usage. The ACN allows agents to talk to each other and negotiate before completing a transaction
- FET token, update of economic schema of FET token costs and incentives
The Simple OEF is the search-and-discovery mechanism for AEAs. It provides a simple HTTP interface for agents to register, describe themselves and then perform searches based on distance or subject. We are building upon it at the start of 2021 by doing the following:
- Identity support: proof-of-identity for peer-to-peer trust
- Integration with main-net: FET tokens are required for complex search and large data operations and give additional incentives for node operators
- Peer-to-peer networking: multi-mode operation, specification and initial work
- Semantic and active search design: design for community feedback of semantic, AI-driven active searches
Q1 is a key period for evolving the SOEF’s current high-performance implementation towards a decentralised digital world in cooperation with development partners and the community: the #metaverse for autonomous agents.
We recently launched our first collective learning field trial focussing on the healthcare industry. Our collective learning network delivers a secure, global AI model to detect COVID-19 and is improving health outcomes. The tool enables distributed parties to work together to train machine learning models without sharing the underlying data. Multiple global participants can train and benefit from the model. The nodes in this experiment were located in Los Angeles and London clusters but new results in one location can benefit all! We hope to see nodes on every continent eventually.
This field trial was just the beginning as we want to improve and leverage our collective learning framework to cultivate a large impact across a number of industries.
To help support this, we are focussing on improving the framework during Q1:
- CoLearn User Interface: Public user interface for interacting with the back-end of the collective learning system
- CoLearn model upgrade: Improved logic for specifying a collective learning model and leaving and joining the model training process
We will be launching toolkits to enable Agents into various applications where business solutions can be built using multi-agent systems. These toolkits are an essential part of making the technology accessible to all. It builds on the agent registry to make it easy to design, test, build and deploy agent solutions that interface with existing real-world technologies, such as messaging, databases, scheduling and more. This builds on the work we’ve done on the decentralized delivery network, hospitality, supply-chains, traffic management and more in recent months.
All of these technology upgrades in Q1 are working together towards the same goals: we want to create a fully integrated system between the ledger and our frameworks that is accessible and makes it as easy as possible for the community to build upon the Fetch.ai network. FET is an essential part of this: it provides the fuel for agents to get things done, the method of value exchange, and the way in which network users operate the network and take part in the decisions that map out how it evolves. The work at the start of 2021 prepares for and guides us through the launch of mainnet version 2.
To keep up-to-date with all our progress and all our other updates, make sure to join the conversation on our socials below.