Fetch.ai Community AMA Recap (6th November 2020)

Nov 10, 2020

Thanks everyone for joining for our Community AMA. In attendance we have:

Humayun Sheikh — Founder+CEO

Toby Simpson — Founder+COO

Jonathan Ward — Chief Technology Officer

We’re going to do a short update on some stuff we’ve been working on, then we’ll do a couple of rounds of those, so that we can cover as much as we can.

Please note we are focusing here on community, ecosystem and tech development. We cannot discuss any price activity, but you already know that.

A: Oh my, where do I start. Agents, agents, everywhere, doing tons of stuff! We’re entering a really exciting phase right now with our DDN (decentralised delivery network) and hospitality work: agents representing these parts of the economy are running 24/7, and you can see them yourself on the OEF if you connect your own agents there. We’re about to transition into field trials for both these projects. We’ll be (lockdowns permitting) using real cars and real people to deliver people, then packages, then food in the coming weeks and months and we’ll see agents booking hotels and representing hotels… (Toby)

A: We have been very busy on Mettalex as you know. The past weeks we have been navigating some regulatory work, following the UK FCA crypto derivatives ruling, but we’re getting to a good place. The good news is we are gearing up to launch an Early Access Program is starting in the next couple of days. The aim here is to recruit active crypto and commodities traders to join the platform in waves over the coming weeks. Stay tuned for that very soon (Humayun)

A: … it’s the really, really cool time when these agents are out there, doing real things, delivering real utility, using FET tokens, the Fetch network, our OEF and our agent framework (Toby)

A: Also MTLX liquidity rewards program is still live til Nov 13th: https://medium.com/mettalex/launching-the-mtlx-liquidity-rewards-program-a1a9046b523f (Matt)

A: Oh, and we’re releasing another version of the Agent Framework today (v0.7.1) which is coupled to yesterday’s release of the OEF 0.2.7. A stack of new features, lots relating to incentivised testnets, and some other fantastic new features for agent discovery (Toby)

A: I’m working on a few things; planning for the next six months and the launch of our v2.0 main-net, plans for incentivized test-nets and the different themes of governance, random beacons, consensus, agent networks, collective learning. Some discussions with potential commercial partners (Jonathan)

A: Any staking of FET for MTLX or Atomix rewards will come from the current staking process (Jonathan)

A: The Early Access Program is launching very soon. We will be selecting a few waves of users to join. Initially we’re focused on professional / experienced traders in the crypto/commodities space. But anyone interested can register. The first wave will be small — 5–10 people — then more from there over November (Matt)

A: Presumably you mean a mobile wallet? We’re working on this at the moment — we have a browser extension and block explorer wallet for interacting through the browser that we’re due to release in the near future (Jonathan)

A: Thanks Josh, we are really interested in this kind of stuff. We have done temperature agents before and we have a how-to for that. And as we continue with the testnet program we’re going to be increasing the amount of documentation for our target use cases (Matt)

A: Yup, those details are coming partially today and then up until it starts on Monday. Our 0.7.1 release today of the agent framework and docs provides a whole pile of the information that you need to get started. We’re super-looking foward to AW2 because it’s not just procedural, it’s real, your agents will do real things, in real places, and can be traded with by others (Toby)

A: Also the OEF stuff is great today, too, it adds positionless searching, more detailed service key searching and some other preparation work for decentralisation. Agent Framework wise, our docker images are updated for all platforms to get people going faster, we’ve reduced memory use by the messaging system by more than an order of magnitude and lots of other great stuff (Toby)

A: Great question. Most things boil down to making things more efficient and cost-effective. This applies to the agent framework in areas like supply chain and collective learning as well. I’m most excited about creating entirely new business models with these different technologies, which is also often what our commercial partners are looking for as well (Jonathan)

A: We are working on a collective learning project focused on healthcare we’re hoping to release a case study on soon. You may have seen Emma’s demo in the summer. This is a research project at this stage, but we’re confident we can produce some value in healthcare setting. Achieving consistent data format to enable ML is a key thing, and where most of the complexity in decentralised learning comes from (Matt)

A: We addressed the plan with staking in this blog: https://medium.com/fetch-ai/staking-and-node-operation-the-path-forward-5ecfc10941db. The plan is future FET network token staking events will be integrated with staking v2 (Matt)

A: It’s not straightforward of course, but there’s no reason why a decentralized alternative to Uber couldn’t be built on the Fetch network. We’re very keen to encourage our community to start building these types of applications on our network. The incentivized test-nets are a great place to start (Jonathan)

A: We are working on a mobility demo, and documentation release for December. There is also a field trial planned for around the same time where we will embed Fetch.ai solution in vehicles. The vision is to create reference implementations, and then the tools so anyone can build with the tech stack (Matt)

A: MoBi has not had a lot of activity recently but there are some great companies in the consortium who we got to know through the work we did there (Jonathan)

A: Oh yes, so we are doing real-world trials in the coming few weeks with the DDN. It’s up, it’s running 24/7 and it’s getting better each and every day. We’re going to be out in cars exercising the apps we have to talk to the agents and giving it a good work-through (Covid lockdown permitting, of course). We’re going to turn all this into a video you can all see, then we’re going to start releasing documentation so that you can see how the protocols work, how you could build for the DDN, and some real code you can use in building your own agents. Plus of course, this is all in the open: build an agent of your own, connect to the OEF, search for DDN agents, and see what’s going on (Toby)

A: We are looking at a number of areas, but one thing that came up in our Community Crowdcast yesterday, is that it’s important to remember that Fetch.ai feels futuristic and there is a world (some way off) which is automated by multi-agent systems transacting, but there is also a lot of ‘low hanging fruit’ which can be improved TODAY. We are working with some fintech and DeFi teams on automated market making, trading bots and the like in the digital economy. We are also working with hotel booking using common web APIs and the big brokers, as well as building and smart city infrastructure — using quite simple agents to do things like optimising parking allocations. So some big companies, some small. All in this area.(Matt)

Image for post

A: We’re working with several different companies but we can disclose any information at the moment. As a general rule of thumb, the larger the company, the longer it takes to make a public announcement but there is a lot of interest in what we’re doing in many different sectors (Jonathan)

Image for post

We’re working with a few now on mobility related things, i.e., it’s not just us who are building these agents, and that’s one of the things that excites me most: the Fetch technology is now mature enough for everyone to build and get things going, which is a big reason why we’re so looking forward to AW2 and AW3 — everyone can build agents, be rewarded for it, but also see your agent in the wild in the digital world talking to others.

It’s a powerful technology stack:

  • The OEF for search and discovery
  • The agent framework for building your agents
  • The agent framework’s peer-to-peer network for negotiations and communications
  • The Fetch V2 ledger testnets for transactions, smart contracts, and more (Toby)

A: There are several, including those on Matts chart, and several others. The majority are either multi-national or located outside the UK (Jonathan)

A: We have projects we are working with across BSC who are looking for random number service, so that is good and we will update when ready. Also, DRB will be a feature of validator activity on Mainnet v2 (H1 2021) and that will provide DRB as a cross-chain service (Matt)

A: Thanks! We’re also very proud of that piece of technology. We’ve only just scratched the surface of what is possible so far. I would expect it to be used on many other chains and DeFi applications. It also fits very well with our strategy of providing agent and AI-based services to many different chains. The interoperability we get from Cosmos’ IBC protocol also enables us to use Fetch tokens wherever we deploy those services (Jonathan)

A: @HumayunSheikh has had to drop out, but he asked me to post this: also — we recently released our white paper on Atomix — https://atomix.ai/ — and we will be launching a similar stake-drop mechanic for ATMX governance tokens before the end of the year. The vision is that future FET network tokens will be accessed via the main staking interface

A: And I think Matt mentioned above, we talked about the flow of the incentivised testnet in yesterday’s crowdcast. We really are making sure there is something for everyone in this: developers, non-developers, node-operators, service providers and more. With the first ones we’re doing lots relating to agents, so it may seem very developer focussed, but AW3 in particular is all about taking another agent and modifying it to do something different. And as many programmers will say, they got started by poking other people’s code to make minor changes. It’s addictive (Toby)

A: I still like to think of the UK as a English-speaking bridge between the US, Europe and the commonwealth countries so I think we have a great opportunity to expand into the US. We’ll be making a concerted effort in that direction in 2021 (Jonathan)

A: That’s another great question! Actually, there’s a field of research known as differential privacy (DP) that can be used to overcome that problem. We’ll be releasing more information on how we’re planning on using DP in the near future (Jonathan)

A: Good question, and I see that Jon has posted an answer whilst I was typing one, so some extra stuff: remember that agents hold their own data, so your agents hold thier data on your devices and computing resources. They are able to communicate, trade, etc., using public key cryptography so only the intended party can see what is sent. Likewise, communications with search and discovery will be encrypted and require signatures, so there’s a lot there that maintains privacy. It is a vital part of what’s going on: we live in a GDPR world and it’s right that people control who can and can’t see their stuff. Things like collective learning, incidentally, are an awesome demonstration of how all this technology can do amazing things: many can contribute to a model, but without having to disclose confidential or private data (toby)

A: We are actively recruiting validators now — if you’d like to hear more… [email protected] Also bear in mind phase 2 of the testnet program is all about network validation and we can onboard you with rewards as well. We’re looking to attract the staking companies, as well as individuals/ community members to create a proper decentralised network (Matt)

A: Initially Stable Coins and at stage 1.5, FET tokens too (Humayun)

A: Liquidity Providers in Mettalex will initially be limited to stablecoins, but we aim to extend that over time (Matt)

A: Mettalex will algorithmically buy-back FET tokens from trading fees on the platform. The level of that will depend on mettalex trading activity, and the policies set by the network governers (MTLX token holders) (Matt)

A: There is the AEA registry where components are being distributed free of charge. But yes you are right, there is no reason why someone couldn’t create an app or agent and sell it for profit (Matt)

A: Short answer: Yes

Longer answer: Still yes 🙂 If you’re creating agents that can attach to APIs out there, then there is a business in delivering those. Even better, there’s one in operating them: people may want agents, but not the pain of hosting them. And, of course, any agent out there that delivers value is paid in FET tokens for delivering that value, so it’s a multi-level economy. And as Matt pointed out, there is a registry for public, free agent components out there which we will see fill out a lot in the coming weeks with AW2/3 in particular as well as our own developments and others. So there are free markets, paid markets and more (Toby)

A: (and longer answer extension: agents that tap into ML/AI in order to be more effective at delivering their value are also attractive to others, so there’s those services behind the scenes powering agents as well as the agents themselves. Plus there are some very interesting strategies for navigating the OEF’s digital world, too, some of which will be more effective than others) (Toby)

A: We’re very keen to provide a solution to the issues raised by the “Social dilemma” — the openness and transparency of blockchain tech is a great way of avoiding a single entity having control over our lives (Jonathan)

A: You can dip in and out of the testnet challenges. It’s not a linear flow (Matt)

A: We’re focused around testnet > mainnet, and a big part of that is pushing increased cross chain compatibility to the mainnet so FET and Fetch.ai services can be delivered across web 3 platforms.