$Fetch_ai AMA Summary with Decentralized Club✔️
Welcome to our community on t.me/decentralized_club Telegram group’s AMA. We really appreciate all your support. There were really good and thoughtful questions that were asked. For those who missed our AMA, this blog post is for you :)
✔️Introduction with Fetch_ai :👉
Q1. Can you tell us a bit about Fetch.AI ?
#fetch: Fetch.ai came up after many, many years of thinking. It is only recently that it has become technically possible, and it is blockchain type technologies (along with other associated cryptographic goodies) that has enabled it.
We often thought that today’s modern life is becoming increasingly more complex, with more and more moving parts. In energy, healthcare, transportation, supply chains, and indeed just about _everything_ these days, there are too many bits and pieces for us to manage effectively. We wanted to bring those parts to life: for them to have autonomy to get things done without human intervention. We realised that with a few additional innovations, we’d be able to create a technology that would provide the scale, intelligence and power to deliver such a vision.
We started working on Fetch.ai over two years ago, looking at the things we’d need to create to deliver potentially billions of these little parts all talking to and doing business with each other.
This is about delivering solutions to you. Directly. Without you having to do anything. It’s about utilising the excess capacity in the economy: using decentralised AI to get more out of what we have, increasing utilisation and frustration, and making the economy dynamic and able to re-organise itself to solve problems instantly. And it’s about hiding the technology, making it transparent, seamless. As Douglas Adams once said, people don’t want technology, they want to get things done. And Fetch is about getting things done.
Q2. What significant developments did Fetch had during 2019?Any other important milestones left to achieve in 2019?
#fetch, Toby Simpson:
You are not wrong! It is also about enabling those things to get themselves done. I know it sounds weird, but we’re surrounded in dumb assets. Shipping containers can’t cry out and say that they’re empty and need filling, but they SHOULD be able to. It’s Fetch and its autonomous agents and AI powered smart connectivity that makes this a reality.
2019 has been an incredible year. We have created vast, vast amounts of high quality, high performance, reliable C++ code driving our smart ledger, our ai-powered smart contracts, our unique PoS with unpermissioned delegation, etc. You can see this on GitHub (https://github.com/fetchai) and also in the wonderful tools we have at places like https://build.fetch.ai — including a transpiler that converts Solidity to Etch (Ethereum Smart Contracts to our Fetch ones).
Then there’s the other stuff. Our partnerships with T-Labs, Grey Swan Digital, the Turkish Steel Mill and many more. Back in February we became the second company in 2019 to hold an IEO on Binance Launchpad. The public sale was hugely successful and we raised $6m in just 22 seconds. Since then, we’ve made huge strides in delivering Fetch (each and every technical milestone ON TIME) and in building partnerships and growing the number of people and companies developing on the Fetch.ai network.
(Just to add: if you’re interested in developing for Fetch, check out https://docs.fetch.ai, and https://build.fetch.ai and join our Slack channel at https://fetch-ai.slack.com/join/shared_invite/enQtNDI2MDYwMjE3OTQwLWY0ZjAyYjM0NGQzNWRhNDMxMzdjYmVhYTE3NDNhNTAyMTE0YWRkY2VmOWRmMGQ3ODM1N2NjOWUwNDExM2U3YjY we’ll make you as welcome as we can, and we ALWAYS want to support people exploring and developing on Fetch)
I guess I should add one last thing to Q2: mainnet is next month, which is a huge milestone for us!
Q3. Which sectors in the crypto industry is FETCH mostly focused on?
#fetch, Toby Simpson:
I was asked a question like that recently, and I said that it was a tough one to answer, because the short answer is “everywhere and everything”. But that isn’t particularly helpful, and it isn’t fair either, because there are areas where Fetch works particularly well right now, like in finance and transportation/mobility, so we are particularly focussing our attention there. Fetch’s unique ability to do AI *on-chain* using ai-powered smart contracts opens the door to a bunch of things that would be not possible due to the computational load on other systems, such as commodity trading with smart market making.
We’re also working in all areas where the community is strong: working to encourage and support developers, and develop example use cases that provide those that can benefit from Fetch.ai with the best possible foundation to build on. Rollout needs to be careful, as a little of everything everywhere never reaches critical mass to achieve anything useful for anybody, so we’re working with commercial partners and developers where Fetch.ai agents and AI-powered smart contracts can make an immediate difference. As the agent population grows, these different areas will combine and deliver other possibilities for the users of the network.
I haul up transportation and mobility one more time because it’s a great example of combining smart-contracts and agents: it’s a complex problem, requiring live, interactive optimisation and has lots of moving parts. Being able to treat businesses like economic lego, combining and reassembling to suit current, real-time needs, is incredibly exciting and it’s something Fetch enables. And it does this whilst returning control and ownership to the individual.
There’s a lot to look forward to.
(in the crypto industry itself, if you separate that out from “industry as a whole”, we’re focussing on what we do that cannot be done elsewhere, such as the computationally heavy smart contracts, ML and AI in smart contracts, the agents and impact in the IoT space, and the performance and reliability of our sharded ledger and how we shine as a performance system)
Yes indeed, and there is a risk to doing everything for everyone: you end up spread thin and achieve nothing. So we are focussing our efforts on areas inside these spaces where we can make a significant difference immediately, and letting applications spread and combine from there. We’re also engaging with commercial partners and the community as a whole to encourage people to experiment, build and try. Developers are super important in this process which is why we’re putting so much effort right now in to making the whole developer onboarding super easy: reducing risk, in particular, for just trying things out.
Q4. How is FETCH planning to handle stringent crypto regulations imposed by the governments everywhere?
#fetch: Good question.
And the answer is this: we handle it by engaging directly with the regulators. They *want* to listen. This is like the Internet in the 90s. People imagined it as a bunch of tubes, or some kind of information superhighway, all of which is fundamentally wrong. If you regulate on the basis of a misunderstanding, it’s going to end in tears for everyone. The best way of making this space work for all of us is to talk directly, communicate, understand the concerns and try and get regulation as good as it can be to allow all of us to operate in a safe space whilst continuing to allow the innovation that we all want to happen.
This, incidentally, is why we, alongside Ripple, NEM, and EMURGO/Cardano, co-founded Blockchain for Europe last year.
Communication is key, because we all want many of the same things: safety, clarity, freedom to innovate.
🐧Community ask Questions by Twitter ✔️
Twitter Question No 1*
Why didn’t you decide to create AI powered devices too along with agents?
#fetch, Nice question. And an easy one to answer: AI powered devices BECOME AI powered devices when you add agents. Agents are the AI: they’re the software units that bring devices to life. Then the Fetch network learns how to connect them all together and ensures that they are best connected to other agents that they are most likely to do business with. We’ve made it super easy to develop them, too. See https://docs.fetch.ai/aea/ for some of our docs on this.
Twitter Question No 2*
What is the current progress of Fetch_ai What is next After reaching 30k TPS?
#fetch, Progress is good! Now we have the beta network up and running with the full unique consensus, that’s a lot easier to stage, test and see. You can, of course, download the code, compile it and run and see yourself, but for others, we’re going to be updating our status pages and block explorer to show these figures for everyone to see.
One thing about this figure: fast transactions don’t exist in isolation. For us, it is because many billions of agents need to get lots of work done and that requires the capability to perform transactions at this kind of rate. Figures like this should never be about saying “my one is bigger than yours” it should be able getting the right performance and right capabilities for the problems that you need to solve. We started with Fetch’s needs, and worked backwards.
Twitter Question No 3*
How easy and fast it is for users of different chains to use FETCH’s ML/AI based contracts?
#fetch, Toby Simpson:
I like this question, because there are two answers. If you’re already familiar with Tensorflow and are confident in the ML space generally, you’re going to find it very easy indeed. ML is a broad subject, as is AI as a whole, but we have gone to great lengths to make it a familiar, welcoming space for those who already work in the ML/AI space. For example, we have a safe fixed point maths library (for auditing and determinism which is vital in MPC), we designed it to support low-poweredd devices, our core ML library supports CNNs, word2vec an even BERT stuff for NLP processing. It’s trivial to set up sequence models, etc. We don’t support RNNs (LSTM) yet, but we will in Q1 2020. The scope for things you can do is vast! There’s some stuff here https://docs.fetch.ai/smart-contracts/pipenv/ and more to come.
Another link for this one — https://docs.fetch.ai/etch-language/ml-overview/
There is a lot there. And for all the ML enthusiasts out there, I do encourage you to look at what we have. The second link in particular. Stuff with the Boston Housing Dataset is one of those old classics that shows how to reimplement existing ML things in Etch on Fetch’s platform, so that’s a cracking place to start.
Twitter Question No 4*
Will the agents in Fetch_ai be cryotographically secured?
#fetch, Interesting question, because it is potentially quite broad and can be interpreted in different ways. It’s a touch like asking “will there be 1s and 0s in the program”, but I think I get the gist of where you’re coming from. Apologies in advance if I miss.
First things first: agents represent what they have. Fetch never stores what they have. Agents advertise specific things built from data models that can be used, with technologies like dimensional reduction, to locate each other (see https://medium.com/fetch-ai/a-decentralised-virtual-world-of-the-future-61d4ecb16b9f for some stuff on this). Once agents find each other, they can transact with encrypted messages that only they can see (i.e., I send you something encrypted using your public key, only you can decrypt it). So that takes care of data transfers…
… but there’s more, of course. We secure the system as a whole with state of the art cryptographic technology and we enable and support things like multi-party computation; a whole field of magic which allows individuals to work on data (or indeed, execute code) without seeing either.
We also support the system with things like verifiable credentials (through integrations with technologies like Sovrin, see https://medium.com/fetch-ai/removing-the-honesty-box-from-the-economy-with-an-anvil-95ca27c6078d — verifiable credentials in a few lines of Python!)
Twitter Question No 5*
How will Fetch.AI solve the problem in cross shard transaction packing? Is there a scenario of multiple conflicting blocks.
#fetch, Given that we’re running out of time and I have a habit of long answers, I’ll be brief on this one. A great question. And we do have a unique solution to doing this which gives us the performance we need. I recommend our paper on this: https://fetch.ai/wp-content/uploads/2019/10/Fetch.AI-Ledger-Yellow-Paper.pdf as it covers this in great detail. Our packing and cross-shard syncronisation mechanism is one of the things that I’m most pelased with.
📓Telegram live Questions answered by Fetch_ai✔️
Q. Can you explain more about fet staking and rewards?
Also see our actual staking site at https://staking.fetch.ai
We have a unique Proof-of-Stake with unpermissioned delegation. This solves a LOT of the issues associated with such things and we’re deeply proud of it. We’re on Auction number 2 (which we ran earlier today) which has locked up 160,000,000 FET tokens!
Q. btw what about mainnet everything is on track?
Ans:Yes. All on track. Release next month. One more code release prior to mainnet.
Q. Can we use all the features of the FET over ERC-20 and BEP2? There should be more benefits on mainnet. Tell us those things!
Ans:We will announce how this will all work with the mainnet roll-out next month, but ERC-20 and Native FET tokens will co-exist. We have some pretty amazing technology that allows the two to exist in harmony with each other and will, of course, release all the code as usual along with the audit reports.
Q. How you will compete with your competitors? Currently i don’t see much but for future how you will maintain this consistency ?
We are solving a different problem: on-chain AI, solving large-scale optimisation problems, and bringing the economy to life with autonomous economic agents. All the structure of Fetch supports all of this. Obviously we compete in different ways technolgy wise, such as our consensus, our ledger, etc. Others have PoS, but our unpermissioned delegation is unique. Others do sharded ledgers, but the way we solve cross-shard communications and integrate our contract execution system is unique. On top of that, being able to do computationally heavy AI and ML in smart contracts is incredible. Coupled with agents, it’s a game changer. But in the end, as I said earlier, nobody cares about technology, they care about getting things done, and we have the ultimate toolkit for doing just that.
Q. Does FET have plans to implement privacy in Fetch? Does IoT currently require a lot of hardware privacy and does applying AI — Agents to control the device cause the device to be deactivated?
Ans:- Agent to agent comms are private.
- Agents never store their data on the ledger or inside Fetch, just representations of a data model for match-making
- Communications can be in private.
- Homomorphic encryption and other technologies when embedded in agents allow for privacy preseving computation to be performed.
- All agents are in control of their value and data at all times.
👬Thanks to all of our Community Fans✔️
Have a Great Day
Edited by : 🌿DK✍️