The Big Interview

Fetch.AI wants to become the ‘Google of autonomous economic agents’

In an exclusive interview, Coin Rivet sat down with the CEO and CTO of the Fetch.AI project to better understand their vision for a smarter future enabled by a host of autonomous economic agents

Fresh off the back of its successful Binance Launchpad token sale, Coin Rivet had the opportunity for an exclusive visit to the Fetch.AI head office to speak with the team about the project’s future goals.

With the project having sold out its token allocation and raising $6 million in just under 10 seconds, the team now looks set to use that capital for their ambitious plan of creating a native blockchain with artificial intelligence capabilities built in to allow users and businesses to better navigate the highly interconnected world around us.

Not just about blockchain

CTO and co-founder Toby Simpson said the project was not just about a blockchain, stating: “Where it gets interesting is when you start combining it with other enabling technology like multi-agent systems, parts of AI, and machine learning.

“You start joining these together and you can start building things that were just simply not possible before. That’s really how Fetch.AI came about, but it’s been 10 years in the making.”

He went on to describe how “people don’t really want technology, but they want to get things done,” and how difficult this sometimes is in our modern life. He claims: “These days, you need five different apps in order to work out how to get from Cambridge to London without finding yourself standing on a random platform waiting for a train.”

Simpson thinks that the capability that the Cambridge-based team are building can help build a system where the various decisions and systems required to make such a trip can be optimised and executed using a network of “autonomous agents” collecting data and making “predictions” enabled by the Fetch.AI protocol.

The CTO then showed off a mobile application that the team has been working on that can access the location-based data from a mobile phone to provide the Fetch.AI network information about the “agent’s” location and direction that could be used to help assess things like traffic to allow other agents to make more informed decisions.

Examples of these decisions include: “Should I leave now to set off on my journey or should I wait another 20 minutes to spend less time sitting in traffic?”

Autonomous economic agents

Toby went on to explain: “If we can get these autonomous agents – or autonomous economic agents as we call them – to run around and give them the authority to be able to solve problems on your behalf, talk to each other and reduce all the friction, then we’d have something really cool.”

Humayun Sheikh, the project’s CEO and an early investor in the now Google-acquired DeepMind, said that we should also “think about decentralisation as a multi-agent system, because what you’re essentially doing is that when you have a decentralised solution, there is no central governance as such.”

He went on to say that in the type of multi-agent system the team are building, “you have various stakeholders, and all these stakeholders are kind of coming to an arrangement to deliver something which is optimal for all.”

A practical example he gave was using this type of approach to potentially optimise the logistical flows of goods that are shipped from a supplier to a final delivery address.

Humayun said many of these networks are currently inefficient due to the hundreds of warehouses filled with inventory and unoptimised delivery routes that could be costing the consumer between 20-30% of the total cost of the goods being procured.

Solving the ‘blockchain trilemma’

The team’s blockchain design to achieve this level of decentralisation on their native chain will use a Proof-of-Stake (PoS) consensus mechanism that incorporates both a decentralised random beacon and also a DAG (Directed Acyclic Graph).

Fetch.AI’s head of research Jonathan Ward thinks this solution can achieve transactional capability “beyond today’s existing systems without compromising significantly on security or decentralisation.”

This solution to the well-known trilemma of trade-offs between scalability, security, and decentralisation for open blockchains has been called a breakthrough by the team, with Humayun saying that the ledger “underpins deployment of ‘multi-agent systems’ where AI agents undertake large numbers of low-value transactions,” like using trading data from a sensor to help make decisions.

Nothing but tools for decision-making

The Fetch.AI CEO went on to say that in his opinion, “AI and machine learning are nothing but tools for decision-making.”

The CEO concluded: “As far as we’re concerned, we’re not building an AI. We are building a framework where you can bring in these predictions for enabling autonomous decisions. When there are millions and billions of agents running around, and also millions and billions of assets wanting to be sold, what you need in the middle is the Google of agents.”

Stay tuned for part two of this exclusive interview coming soon on Coin Rivet.

Related Articles