Understanding the Fetch.ai network: a guide

Feb 8, 2019

There are many unique and innovative aspects to the technology we’re developing. Fetch.AI isn’t just ‘another blockchain’ for recording transactions. It’s also not just a machine learning system for making predictions. Rather, these technologies are part of a broader goal. Our team is combining them in new ways to create a web infrastructure that delivers on the potential of Autonomous Economic Agents (AEAs) for the first time.

For the uninitiated (where have you been?!) agents are simple pieces of software code that can plan and reason about the world to solve problems autonomously. Agents can live on machines of all types (cars, laboratory equipment, sensors) or represent real-world assets (hotel rooms, airline seats, meeting rooms) helping them deliver economic value (as journalist Nick Booth outlines well here).

With this in mind, we wanted to detail the new capabilities that the Fetch.AI network makes possible. There are five core functions agents need to autonomously trade with each other.

Search and Discovery (advertising services) — An agent that has goods or a service to offer needs to be matched with an agent that may wish to purchase it. The Fetch.AI network is designed to allow agents to advertise their services, broadcasting their offers far and wide to discover potential customers. Similarly, the use of machine learning technology delivers increasing market intelligence over time, placing agents that are likely to do business close together in the network.

Communication — Once agents are ‘matched’, a potential economic exchange can take place. For this to occur, agents must speak the same language. Without a common language it would be impossible for agents to negotiate. Imagine a Russian-speaking chef attempting to negotiate with an Italian-speaking market stall owner. That’s why we’ve incorporated a Common Dialogue Framework into the system.

Negotiation (agreeing the price) — agents need a common set of rules and a mechanism by which they can reach a fair price for a given exchange. The Fetch.AI system is designed so that individual agents can deploy a wide range of strategies to achieve the desired outcome for their creator. We’ve also developed a set of standard algorithms for generic negotiation strategies to help the community begin building agents.

Collaboration — While agents have different goals i.e. they negotiate to get the best price or to achieve their objective, they can also collaborate for their mutual benefit. The ability of Fetch.AI to enable agents (and therefore their owners, machines or stakeholders in the real world) to cooperate is one of our main points of value. We’ve outlined the role of Synergetic Smart Contracts by describing how they could help taxi drivers to collaborate in order to achieve economic gain for all participants.

Execution & Trust (trading) — Just like you or I, agents need to be able to undertake economic exchanges, safe in the knowledge the terms they have negotiated will be securely upheld. That’s where the Fetch.AI Smart Ledger comes in, providing a highly scalable, near-zero cost record of the ‘global state’, accurately accounting for value transfer across the network. Similarly, the Open Economic Framework can also provide trust estimation for different agents over time, helping to incentivise and ensure good behaviour to remove counter-party risk.

With the key building blocks in place, the stage is set for autonomous agents to begin solving a whole host of problems autonomously and efficiently. Whether it’s enabling autonomous vehicle ride hailing and pooling, or more efficient parcel delivery optimisation, we believe Fetch.AI is a decentralised network with all the components that the world needs.