23. July, 2019

Fetch.AI AMA with CTO Toby Simpson

Thanks to everyone who participated in Fetch.AI’s AMA on 11 July on YouTube. We received an enormous number of questions and unfortunately we ran out of time to answer many of them. If you missed the live AMA, you can watch it in full below:

In this article we’ve provided answers to 10 of the best questions discussed during the AMA with our chief technology officer Toby Simpson. At the bottom of the page, we have also added a further three questions. Although we didn’t have time to address these during the AMA, we believe the answers will be of interest to the Fetch.AI community.

I read the terms ‘Universal Integration Protocol’ and ‘friction-free integration for siloed systems’ on the Fetch.AI website. What do they mean?

The internet allows us to talk to one another regardless of where we actually are. By enabling Fetch.AI’s autonomous economic agents to interact, we realised we could enable communication and negotiation on a grander economic internet level.

One of the problems that we have in the economy today is a lot of the information that we use on a daily basis exists in its own individual silo. These silos are often unable to communicate with each other. When they do, they tend to do so in a proprietary way with different protocols for connecting them together. There isn’t a mechanism where all of them can work together. That’s where we step in. Fetch.AI allows you to take apart the capabilities from one system and join them together with capabilities of another system to create something new in real time.

What do you think will be the greatest benefit of Fetch.AI’s technology?

I guess the greatest benefit is that Fetch.AI will make our complicated lives much simpler. We have access to all sorts of information and knowledge, but the problem is that you as an individual are required to trawl through multiple different applications to coordinate everything. This problem will only worsen as more and more data becomes available.

Decentralization provides us with scalability and also personal ownership. This is a big step forwards in how it is that we will conduct our lives in the future. I think that all of the work that’s going on in this space is really exciting because it’s giving individuals power over their own identity and the claims made against it.

How safe is Fetch.AI?

I presume you’re not asking if the technology will leap out of your computer and steal your pets in the middle of the night. I guess you’re asking about privacy and security. At the moment if you set up accounts in two separate banks, you have to provide all of your personal information twice. Providing data in this way is unnecessarily time consuming and complex.

Fetch.AI uses verifiable credentials based on zero knowledge proofs and this ensures the data is under your control and prevents it from being shared with others through digital leakage.

We’re often asked whether, if our autonomous agents are running around representing things, does all of the data sit on a public network? Let’s address this clearly. Firstly, autonomous agents do not represent things on the network. It’s only when agents are introduced to each other and they wish to perform a transaction that any data is actually exchanged. In these instances the data can be encrypted and it can be exchanged directly between the two parties. Some of you may also be aware of privacy preserving computation which is one of these neat tricks where you can do computation without knowing about the data that you are working on. In addition to this, we are also using other technology such as homomorphic encryption, which allows people to perform computation without having access to the individual bits of data.

Ultimately, when you own your identity, you own your utility.

In simple terms, how does Fetch.AI’s Open Economic Framework (OEF) function?

One of the things I love most about Fetch.AI is the OEF. It’s the interface to the world for our autonomous agents. It gives them the ability to search for each other and interact to gain information. In the real world we are all constrained by the reality that we are in. For example, all the people in this office all see the same office, albeit from a marginally different perspective. In the digital world, no such restriction exists and it’s possible for everybody to see a view of that world that’s uniquely optimized for them. In the digital world, a meter doesn’t have to be a meter as you can restructure the world in order to suit the individual who’s observing it.

In this way, autonomous agents can see in front of them what is most relevant to you at that exact moment. The OEF has been designed to enable semantic teleportation, which is a cool sounding name for positioning yourself next to people who want the same thing. This means your autonomous agent can explore other opportunities that you may not have considered. By doing so, the agent is able to deliver more value and find more of the things you seek.

The OEF serves an introduction agency. Therefore, regardless of what you have, or whatever it is that you want, your agent is introduced to other agents that are most applicable to you at that particular point in time. This can be achieved on many different levels: by content, by geographic location etc. For instance, in one city you can look towards another city and see everything related to transport.

Will Fetch.AI consider partnerships with the tech giants in the space such as Grab and Airbnb?

We’re working very hard and we welcome collaborations with anybody who’s interested in a new ways of building things on the Fetch.AI network. We are focusing on the underlying technology of the smart ledger and all of the other bits and pieces that are required in order to make Fetch.AI work. We’re also helping people develop applications and their ideas on the Fetch.AI network. We have provided a huge amount of documentation on our community website and, of course, a vast quantity of code on GitHub. We have also got a developer Slack channel where we can talk and support people. We’re always interested in hearing people’s feedback and ideas.

How did you come up with the idea of ‘My Easy Charge’ and what change does this development bring to the real world?

‘My Easy Charge’ won the Trusted IoT Alliance Smart E-Mobility Challenge at Bosch Connected World 2019 in Berlin. We developed the technology in collaboration with Bosch, T-Labs and Share&Charge.

We believe it is a really good example of our unique Synergetic Contracts. You have lots of electric vehicles, a whole bunch of charging points and a period of time that is required for charging to take place. The technology takes into account relevant contextual information about each individual’s preferences. For example, you may not mind waiting 30 minutes to charge your vehicle, if you also happen to be starving and fancy a spot of lunch en route to your destination. You can watch the video discussing the use case in more detail here:

We believe aspects of this technology can be incorporated into the broader context of the energy industry. This is something we are very excited about.

Fetch.AI agents could potentially be used for trading and an example would be that of crypto market makers. But would it be feasible to use an agent for exchange-performing statistical arbitrage and communicating with each other for interexchange arbitration?

The answer is yes, of course! We’re working on some very interesting things on this subject area, which we will be able to talk about very soon. You’re right though, Fetch.AI’s autonomous agents are an ideal way of trading. When you combine agents with Synergetic Contracts then you end up with something extremely interesting that has the ability to perform the necessary heavy computing.

What inspired the founders to create Fetch.AI?

I come from a computer game background. I have experience of building large-scale virtual worlds using agent-based technology and biologically inspired artificial intelligence. Before we started Fetch.AI, CEO and co-founder Humayun Sheikh and I speculated on and off for quite some about whether or not it might be possible to represent a larger proportion of the economy using autonomous agents in such an environment. We hoped to be able to customize it in order to connect people together so that they could get work done. Decentralized ledger technology has provided some of the necessary enabling components. It’s given us scale and computing power and when combined with other technology, we realized that what we had dreamt of doing with Fetch.AI was possible.

Can you explain the difference between Fetch.AI and other AI technology? What makes Fetch.AI better than other AI?

There are many ways of stacking cats, but one of the things that we wanted to set out to achieve with Fetch.AI is collective intelligence. All of these individual bits of knowledge and intelligence in existence can be shared to the benefit of all. Not only does our technology enable the deployment of Synergetic Contracts, it provides knowledge services and optimization services to others too. It also allows you to interpret what is going on in the network in order to make accurate predictions.

Of course, there is the AI that exists on the network itself. This powers how the OEF learns how to introduce people effectively. It’s also the stuff that powers reputation and trusts systems.

What happened to Useful Proof of Work (µPoW™) on page 16 of your whitepaper? I haven’t heard much news on that one…

It hasn’t gone anywhere! µPoW™ is still in the system. We have generalized the concept quite extensively and the first iteration of this is Synergetic Contracts. These contracts are genuinely useful. They are economically useful to the contract itself and there is all sorts of exciting stuff relating to how the rewards for that can potentially be used to be delegated and staked on your own contract’s behalf. There is a lot more to come on all of this, but µPoW™ is something that’s delivering far, far more than we imagined when we first started talking about the concept, all those years ago.


Additional questions not answered during the AMA:

How did you guys survive in this long bear market and what difficulties have you encountered during this period?

The ‘crypto-winter’, as it was also referred to, is not unique to this space. We survived because we have had, and continue to have, great support and great technology. We’re trying to do something different, where there is a clear market need and we have a fantastic team delivering it. In any market, support and contingency are important: you need the time to be able to get to, and beyond, each key milestone to demonstrate the opportunities available.

With the many tokens that exist today, not to mention the emergence of Facebook’s Libra Coin, what are your plans for Fetch.AI on the world crypto market?

It’s an exciting time. As I often say, it feels similar to the mid to late 1990s when the Internet and the web were just developing. Just as was the case then, we now have a game-changing set of technologies, and some incredible exploration into how it can be used to solve existing problems. This will help us to deliver benefits we have never previously imagined. Libra is a strange one, but it does bring crypto out into the open and lets us all look at raising our game to ensure our technology is easy to use and integrate into people’s lives. At Fetch.AI, we continue to focus on our unique combination of AI, crypto, multi-agent systems and more.

It looks like you guys are going to build your own mainnet. What’s your reasoning behind not using a current solution, and using plasma or another high throughput chain? Is there a reason that Fetch.AI needs to build one out that’s entirely new?

Different problem, different solution. The Fetch.AI network requires billions of autonomous agents to connect together in real-time. The need to do this, and to enable the agents to learn who should connect to who, required a new set of technologies that were not available when we started out. Indeed, most are not available even now.

To build the network, we needed on-ledger AI (including machine learning), we needed to support low-cost, high-throughput transactions (for many agents) and we needed the underlying information and structure to deliver our decentralized digital environment for agents. These are the key “family jewels” which have allowed us to get the network performance, scalability and robustness we desired. Crucially, it also ensured we could easily integrate our innovations in smart contract technology and consensus.


St. John's Innovation Centre,
Cowley Road,

Go to map
Technical: [email protected]
Investors: [email protected]