In this episode of the podcast, we welcomed Andrej, CEO and co-founder of Wynd Labs, to share his experiences in developing the Grass network and the Live Context Retrieval (LCR) technology. Andrej discussed his transition from a PhD in applied mathematics to entrepreneurship and delved into how LCR technology is transforming the way AI models access real-time data, overcoming the limitations of traditional AI, which relies solely on static information.
We also talked about how the Grass network ensures data accuracy and transparency, and explored the privacy protection measures within decentralized networks. Andrej elaborated on the technical and ethical advantages of Grass over conventional centralized platforms, and its wide-ranging applications across industries — from e-commerce to financial market predictions. Additionally, Andrej revealed Grass’s upcoming developments in the next few months, including the launch of an Android app, new hardware devices, and incentive programs for contributors.
This transcription was generated by GPT and may contain errors. Listen to the full podcast:
YouTube: https://youtu.be/rFU9zOeWt30
Andrej’s Background and Project Overview
Prince:
Today, we welcome Andrej, CEO and co-founder of Wynd Labs. Can you give us a brief introduction about yourself, how you got involved in this project, and what this project is about?
Andrej:
My name is Andrej. Before this, I was working on a PhD in applied math, specifically in computational physics. After the pandemic started, I used that time to start a company focused on web scraping infrastructure. We set up large servers and data centers for customers to scrape the web. As our customer base grew beyond 1,000, they asked if we had solutions for using residential devices and networks for scraping. I later learned that major companies like Walmart were using real people’s devices to scrape the web by sneaking SDKs into free applications.
I was unhappy with this due to the ethical issues involved. I then met my two co-founders, and we decided to address this problem. We didn’t anticipate the explosion of the field due to AI development.
Now you know that the scope of our project has increased tenfold since we started. Grass is a network that anyone can join with one click. By joining, you’re essentially renting out some of your resources, like CPU and bandwidth, to scrape the public web at scale.
Explaining Live Context Retrieval (LCR)
Prince:
So I think not all our audience have the engineering or computing background. You know, not all of them are familiar with AI technology. So, could you give a very simple explanation about your core technology, Live Context Retrieval (LCR), and how this technology changes the way AI models interact with data? Maybe compare it to traditional data training methods.
Andrej:
Yeah, definitely. To explain LCR, imagine you’re having a conversation with someone, and you bring up a topic that they aren’t fully updated on. Maybe they haven’t read the latest news or something like that. Then, instead of giving you outdated information, that person looks up the latest details on their phone and gives you a more accurate response. That’s basically what LCR does, but for AI models.
In simple terms, LCR allows AI models to access and use real-time information from external sources like the internet when generating responses or making decisions. This means the AI model isn’t limited to what it learned during its initial training. Typically, when you train a model, you feed it a lot of static data, so if you ask it a question about something that happened today or recently, it can’t provide an answer. But with LCR, it’s like connecting an engine to the AI model that gives it access to the public internet, allowing it to answer questions based on current information. The models are powered by millions of nodes that participate in the network, and LCR runs across all of them.
Comparison with Traditional AI Models
Prince:
Yeah, because, you know, as far as I know, OpenAI doesn’t have access to instant information online. And they have some level of human intervention when training AI models, right? They relabel everything that the model responds to. So I don’t know if it’s similar with your system, or if there’s no human intervention at all?
Andrej:
Yeah, there’s no human intervention or tampering with the network. It doesn’t rely on any human response. It’s 100% automated. Moreover, everything will be 100% verified on a public ledger. This guarantees that none of the responses going through the Grass network are biased or tampered with by the network.
Information Integrity and Privacy Issues
Prince:
So have you ever been concerned about the accuracy of the information online? Like, is there a risk that the AI might misinterpret or return violent content, or even just answer a silly question inappropriately? Does your team do anything to prevent these kinds of accidents or problems?
Andrej:
Yeah, so every user of the Grass network undergoes a KYB (Know Your Business) process and must follow certain compliance procedures. When I say user, I’m referring to the customers on the other side — the AI models.
As for the data being accessed, AI models that want to access the internet are currently limited to using search engine results. The problem with search engine results is that they are filled with advertisements, and the information that appears on the first page is often good for humans but not for large language models (LLMs). It’s quite poor quality. There are studies published by OpenAI showing that, in many cases, 40% of queries use only 1% of the available data. This demonstrates how much of the data is not valuable.
The really exciting thing about LCR is that because all the nodes verify every transaction in real time, we can ensure that none of the data returned to the model is optimized for anything other than knowledge and semantic similarity. That’s one of the biggest key differentiators for this network.
Prince:
Yeah, I think that’s a great perspective. So let’s talk more about LCR. How does LCR handle privacy concerns when retrieving live data? Is there any risk to user privacy?
Andrej:
Yeah, there’s no risk at all. Grass nodes have nothing to do with user data or browsing history or anything like that. The node is just using a little bit of CPU and bandwidth. The only data being accessed isn’t on the person’s computer; it’s from the internet. The person’s computer acts as a node, simply routing the web request. So there are no personal data concerns. To further that point, we’ve completed audits with three different groups, and one of those is public on our website. We also completed a review with Apple’s team, which is the leading cybersecurity platform for these types of applications.
Impact of LCR Across Industries
Prince:
Yeah, so tell me more about the major use cases for real-time data. In what ways can LCR benefit industries outside of crypto, like in financial markets or predictive modeling, or any other examples?
Andrej:
In terms of real-time data access, every Fortune 500 company is going to be using AI in some capacity in the coming years, and every use case requires the most recent information. For example, if you’re an airline company, you need a real-time understanding of the supply chain, and you’ll need global data to make certain decisions beyond travel.
If you’re an e-commerce company like Target or Amazon, you need the ability to scrape all your competitors’ websites and know their prices in real time. For instance, Costco knows the price of every single item on Amazon every day. In the future, LLMs and AI will need access to platforms like LCR to do this efficiently.
The use cases are industry-agnostic because, in the next few years, every industry using AI will need access to real-time data. As for predictive analytics in financial markets, that’s another massive application of LCR. In fact, several leading hedge funds have approached us to participate in the closed alpha for LCR.
Competitive Advantages of Grass
Prince:
Yeah, I think there are a lot of teams doing similar work. So compared to competitors, what do you think are Grass’s biggest advantages?
Andrej:
When it comes to competitors in the crypto space, I don’t think there’s any other solution that can do what we’re doing.
Prince:
Yeah, I was wondering, are there any competitors right now?
Andrej:
In the crypto space? Not that I’m aware of. I expect some protocols to recognize the importance of what Grass is doing and try to emulate it, but so far, no one has been successful.
Outside of crypto, our biggest competitors are two companies that can crawl the entire internet. I won’t name them, but they are very big, and I think listeners can guess who they are. Grass has two key advantages:
First, on the ethical side, Grass is owned by its users. Every decision is made by users, and it’s fully democratized. Everything is recorded on an immutable public ledger, so people can verify things for themselves rather than just trusting a centralized company’s word.
Second, on the technical side, Grass can do things that other companies can’t because it’s powered by millions of nodes. It can access websites faster and in a more unbiased way. Since Grass doesn’t have the overhead of a centralized company, it can monetize the network more efficiently. We prioritize page rankings based on how valuable the information is for an LLM, not for advertising purposes. This is a huge advantage because when an LLM requests data, it cares about finding the most honest and truthful information, not which website has the best ads. That’s something LCR can provide that no other solution can match right now.
Prince:
How do you evaluate the honesty of this information? Is there some sort of standard or computing technology behind it?
Andrej:
It’s pure math. What we’re doing is crawling the web with 2 million nodes right now, and the amount of data we’re crawling is enough to train ChatGPT on a daily basis. It’s a lot of data. With 50 to 100 million nodes, we’ll be able to crawl the entire internet every day, building knowledge graphs of all the different sources. We use similarity metrics, which are mathematical equations, to evaluate which websites will provide the best response to any prompt, regardless of factors like advertising. That’s what we refer to as the “honesty metric.”
Incentives for Contributors
Prince:
Yeah, because I don’t know anything about mathematics, but I think it’s very important for me as a customer to know I’m not being manipulated, right? So let’s talk about the incentives. What kind of incentives do you offer to your contributors, and how are users encouraged to participate or validate the web data?
Andrej:
Yeah, that’s a great question. Right now, it’s a fully passive network. All users have to do is install a node, and it does everything for them in the background. However, we’re evaluating ways to encourage more active participation because, even though the application is passive, our community is very active. We’ll be putting out more consumer-facing applications soon, and I think the first one will be using LCR. So keep an eye out for that.
In terms of incentives, up until now, we’ve been focused on building the network and ensuring it’s robust. Having millions of users join early has been incredible because it allowed us to run extensive stress tests and ensure the network works efficiently. As a reward, early participants are receiving an airdrop. But, honestly, the airdrop isn’t the main reason to download Grass — it’s more of a reward for being an early adopter.
In the long term, what really excites us is enabling bandwidth monetization. For example, when a large company uses LCR, if your bandwidth is being used to query the data that company needs, you’ll actually earn rewards for that. That’s the exciting part because it creates a constant stream of passive income for users who simply have the system running in the background without needing to worry about it.
Upcoming Developments
Prince:
What kind of rewards can users expect? Could you elaborate more on that? And is there any news about the next few months or this year? What major milestones are coming that you’d like to share with our audience?
Andrej:
Absolutely. There will be a Grass Android app. A lot of people already know about the Saga app, the Solana phone, and now the Android app will be launched after the network goes live. We’re also working on a hardware device for consumers. The first LCR product, or the first product that uses LCR, will allow Grass users to beta test it. So if you’ve installed Grass, you’ll have the opportunity to try out the first application that uses LCR. There are a few other things as well, but keep an eye out for those developments.
Definitely. One thing I want to say to the audience is, if you’re a Grass user, thank you so much for joining the network. We’re immensely grateful for every single person that has joined because without the people, the network is nothing. And if you’re not a Grass user, we have an amazing community, and we will happily welcome anyone with open arms.
Also, keep an eye out for some of the roadmap items I mentioned, like the hardware device and some of the other applications that are coming soon.
Follow us
Twitter: https://twitter.com/WuBlockchain
Telegram: https://t.me/wublockchainenglish
