Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
BNB AI Hack Workshop: From Data to Insights: AI-Powered Intelligence and Secure Labeling
Recorded: March 27, 2025 Duration: 0:48:55Space Recording
Short Summary
The BNB AI Hack workshop unveiled several innovative projects, including Alaya AI's gamified data labeling platform, DING's infrastructure for decentralized AI agents, and Tagger's solution for specialized data management. These initiatives highlight significant trends in token economies and strategic partnerships within the blockchain ecosystem.
Full Transcription
Hi everyone, welcome to the BNB AI Hack workshop.
Today we are diving into From Data to Insight, AI-powered intelligence and secure labeling.
And today we are thrilled to have three speakers joining us today.
First we have Richard, Head of Ecosystem from Alaya AI.
Second we have Callan, Product Manager at DIMM.
And third we have Trevor, Founder of Taker. So without further
ado let's welcome our first speaker Richard from Align AI. Hi Richard. Hi Tiana, can you hear me?
Yes and the stage is yours. Awesome thank you very much. Thanks for the introduction, Tiana. And hello everyone.
So it's an honor to be invited to the EMB AI Hack workshop again. So I'm Richard,
head of ecosystem ads, Ally AI. So before we dive into today's topic, allow me to provide a brief introduction of Allio AI for users who may not be
familiar with our projects. So Allio AI is an open composable Web3 AI data platform that allows users
to monetize their spare time and earn rewards by labeling AI data through a simple gamified
interface.
Users can earn a variety of rewards, including ETH, BNB, stablecoins, and our native token,
AGT, through flexible participation in various data contribution activities.
participation in various data contribution activities.
So AI is designed to leverage the inherent user acquisition
advantages and network effects of Web3 ecosystems
to supercharge the AI industry by offering the best of Web2
and Web3.
So we have three main modules, a distributed AI data ecosystem
where enterprise data services meet decentralized Web3 data contributor communities,
an open data platform designed to support a vibrant tokenized data economy for early stage and small medium AI developers through customizable datasets bootstrapping.
So we'll get into this a little bit more deeper
in subsequent sections.
And finally, an AI data auto-labeling tool set
designed to supercharge decentralized AI projects
for powerful pre-processing and automation features
in affordable and customizable packages.
So moving on to today's topic, let's explore some core challenges in the AI data sector
that blockchain technology would potentially address, in our opinion.
So we all know that AI consumes data, but for the average average user it is difficult to trace who is collecting their data
how their data is collected and how the data is being used. So under traditional big tech models
user data is generally harvested and privatized by corporations as a default part of the service
package. Users are generally unable to assert ownership
and exercise control over how their data is used,
nor are users able to benefit from data contribution
without forfeiting most, if not all,
data privacy rights to tech giants.
Nor AI models rely on data in one way or another
at certain stages of training and development.
So what happens is as models get bigger and more data hungry,
supply shortages in data quantity and quality
becomes increasingly apparent across various AI applications and sectors.
So for example, OpenAI success with the GPT series relied heavily on strategies like
reinforcement learning from human feedback, which is a model fine-tuning method that is
expensive, labor-intensive, and relatively difficult to scale in centralized settings.
So what happens is that this creates a situation
where big tech monopolies essentially freeload on user data
to profits of data network effects.
This reduces innovation incentives and also erodes public trust
in centralized institutions.
So moving on, let's explore the idea of token-driven data economies
and how they could address these challenges.
So a core advantage of blockchain technology is that it allows the creation of a global marketplace for data.
For data labeling through the use of cryptocurrency wallets, enabling instant payments free from traditional restraints
such as administrative costs and cross-border payment fees.
So with blockchain technology, we have a chance to really democratize AI and data labeling,
giving control and rewards back to the people who actually create and curate data by tokenizing data as an asset.
For example, users can pull private data such as browsing records and then rent them out
as encrypted bundles on a per access basis. So this is the data monetization model that several AI data blockchain platforms are already working towards.
On the other hand, users can also collectively contribute via active data labeling
through decentralized open data markets with transparent pricing features,
such as through custom data pools on Alaya AI's open data platform.
A key advantage of the open data platform is that it can serve both large and small data requests simultaneously,
which enables better horizontal scaling across different data verticals.
Poken-driven data economies also provide alternative possibilities for data validation and quality assurance.
For example, by combining historical data contribution records of individual wallets with on-chain staking and reputation mechanisms.
Flexible governance mechanisms can also be utilized to build robust cross-validation networks to incentivize consensus validation,
which allows complex data tasks to be processed fully on-chain, purely through Web3 incentives alone.
So to summarize, token-driven data economies make it possible for us to create novel economic structures
where users and platforms can collectively share responsibilities and benefits from
data collection and labeling.
This brings us one step closer to building a fair global marketplace for AI, data and
intelligence. Finally, let's explore how blockchain and related privacy technologies can provide us
with a framework for secure data sharing and actionable markets intelligence.
Blockchain technology enables us to create an immutable audit trail, making it possible for users to regain control and monetize their private data
through various encryption techniques and controlled access channels in the transparent markets.
That provides equal bargaining power between data contributors and data consumers.
So let's consider the possibility of creating a public open-sourced AI application that trains on sensitive data without actually seeing the data itself.
hesitant to share data. For example, such as trading and financial research, or in any sort
of area where digging for alpha in crypto technology. That's a really funny mistake.
But yeah. So anyway, it's actually possible to create a data platform, in our opinion,
that allows AI models to be trained on private data without actually revealing the data,
while simultaneously ensuring that data providers are fairly awarded.
So for a trader or investor, imagine the edge that you will get if you have insights that others don't.
I mean, of course, this includes potential risks
such as, I would say, criminal activity, obviously,
which is rapidly becoming a significant sector
or at least like a focus or emphasis by public governance and regulatory entities
across the world.
But on the other hand, so Web3 is also enabling alternative data streams that were previously
untapped or underutilized.
So while it might actually sound quite counterintuitive,
the inherent transparency of blockchain technologies
can also be combined with encryption techniques
like zero-knowledge proofs and confidential computing
to enhance data privacy and security.
So federated learning can also be incorporated
for local data processing, which allows users
to train AI models without their data ever leaving their private devices.
So this could be especially useful in situations where user identity or a sort of sense of
privacy elements have to be carefully handled.
For example, medical data.
And on the other hand, users can also directly feed novel data metrics
into decentralized AI networks to receive real-time market intelligence
from on-chain data sources or market indicators.
So this could, of course, create both a lot of opportunities and also potentially generate risk factors
or potential attack factors for malicious actors
such as the Lazarus group.
But obviously, even if an AI were to be able to identify malicious actors such as the Lazarus group.
But obviously, even if an AI were to be able to identify
certain tech vectors, that would be...
It really depends on how AI technology is deployed
and aligned in general.
But this is still another topic for another time.
But anyway, what happens when we
combine market intelligence from on-chain data sources is that we could create streams of market
signals which can then be directly fed to custom AI trading agents with pre-configured execution parameters or decentralized data streams.
So this allows profitable trades to be further utilized for AI agent training or even profit
sharing, although this is context dependent. For example, through tokenized AI models or shared revenue pools.
So the next effect is that these technologies or combinations
enable market intelligence to become more real-time,
more granular, and also more accessible
without unintentionally, quote, leaking alpha based on crypto degen terminology for market participants.
And also, on the other hand, from a venture capital perspective, this also means that signal detection can be improved.
AI models that can analyze decentralized data might be able to flag emerging trends or products faster than individuals
who are manually sifting through entire walls of threats.
So in conclusion, at AILAR AI,
we believe that the future of AI and data
is technically inseparable from blockchain technology
as there's still a lot of benefits that AI can explore
from novel means of data ownership
and also economic structures that are made possible
through the design and application
of decentralized blockchain ecosystems.
And we believe that this new ecosystem
will be designed around empowering every participant
by ensuring data contributors are properly incentivized and fairly rewarded
and to enable the creation of collaborative AI intelligence ecosystems
where data can be directly transformed into actionable market insights
and where individual incentives are hopefully aligned inherently with the other good.
So this wraps up my presentation for today and thank you very much.
Thanks Richard for the insightful presentation and everyone do check out Liar AI on X and website for the latest update
and thank you again for the presentation thank you very much Tiana and next let's
welcome Kellen from the email hi hi Tiana Hi everyone, this is Callan, the Product Manager at DIN. It's great to be here with all of you and share your ideas about AI. AI is changing the game, right? But there is still one big question handing over us.
How do we make AI more autonomous, scalable, and decentralized?
And that's where DING, Data Intelligence Network, comes in.
A real solution to provide the infrastructure that AI agents and decentralized AI apps need to thrive.
So why DING is focusing on the AI agents is simple
because the future is heading straight in that direction.
And the numbers speak for themselves.
You can see the global AI agent market is set to explode
growing from around 4 billion to like 140 billion by 2033.
And that's not just growth, it's a massive shift in how AI will operate in our daily lives. And the market for AI agent driven cryptocurrencies is also
booming. Right now it's already at around 10 billion and the trading volume alone is pushing
around 2 billion and they both have a very high
growth rate these numbers tell us one thing like ai agents and just a trend they are the future of
decentralized intelligence and ding is here to build the infrastructure to support their future
so now is the here is the key question, like what powers AI agents?
I think the answer is simple, it's data.
So let's go through the AI agents.
We can see they start with data.
Every intelligent system from chat GPT to GROC,
from a fake Elon Musk chat board
to your bank's virtual assistant, is only as good as the data.
What really defines them is the data they've learned from. From the conversation, the labeled
datasets, and the patterns they pick up over time.
And to train these agents, it's a long process.
It starts with raw data, then goes through human labeling teams who refine it.
And finally, the AI models use this labeled data to learn, adapt, and improve.
Without quality data, even the most advanced AI agents
could be useless.
And that's why we put data at the core of our AI agent
infrastructure, because bad data means smarter, more
effective AI.
But here is the catch.
Getting AI agents to learn from data isn't instant before data can be used for training.
It has to go through a long and tedious process.
It starts with gathering raw data, then transforming and cleaning it, followed by fight training,
training and constant evaluation.
It's not just about feeding AI data.
It's about making sure data is structured, high quality,
and meaningful.
When we ask any data scientist, they
will tell you the same thing.
80% of their time is spent on data preprocessing,
while only 20% is spent on actual analysis and model
building.
And that's exactly why we're focusing
on solving this challenge.
By providing efficient solutions for AI data processing,
we will help AI agents learn faster, better, and at scale.
Another reality is good data is very expensive.
And big companies, they know the value of high-quality data,
and they are willing to pay for it.
Google struck a $60 million deal with Reddit for AI training
data.
Stack Overflow is licensed as data to open AI and Google.
Even Twitter has locked down its API,
charging around $100 per month for basic access and up to $40,000 for enterprise plans.
But here's the problem. While these companies are making millions, as data owners, we never get paid.
The content we create, the conversations we have, and the knowledge we share, it offers AI models, yet we see none of the profit.
This is exactly why DIN is building decentralized solutions so that the contributors can actually
benefit from the AI economy. So what do AI agents actually need to operate effectively? We can
break it down. First and foremost, agents need access to decentralized
data with a fair monetization mechanism. Right now, the ecosystems out there don't
offer a system where users can both contribute their data and earn from it. We aim to change
that by creating an environment where data ownership is respected and data contributors are rewarded.
AI agents also need seamless workflows across both on-chain and off-chain environments.
The current blockchain infrastructure, they just isn't built to support autonomous agents working together across different systems in an efficient way.
Another critical need is secure and verifiable computation. anonymous agents working together across different systems in an efficient way.
Another critical need is secure and verifiable computation.
Their agents require trusted execution environments to ensure that computations are performed securely and transparently.
Right now, off-chain execution is limited in terms of trust and transparency which prevents sensitive applications from fully adopting
ai like the id products or like defy products and lastly ai agents need multi-aging systems
the ability to collaborate and even compete with each other the current infrastructure they Current infrastructure lacks proper support for scalable and decentralized mass,
limiting the potential for AI agents to work together in complex and dynamic ways.
Ensure the AI agents urgently need the right infrastructure, and we are here to provide it.
How does DIN solve these challenges?
DIN is here to address all the problems we mentioned before.
We are the AI native data layer for the BNB ecosystem,
building a trustless distributed network that enables anyone to contribute data
and share it in the income it generates.
We integrate decentralized solutions that allow AI agents to work across all environments
easily and fluidly. So with DIN, we eliminate the risks of centralized control and ensure
that data pricing is fair and transparent. We create an ecosystem where data isn't just
a comedy controlled by a few players. It's something that anyone can participate in, contribute to, and benefit from.
Our vision has already gained support from many prominent investors, including Binance Labs, HashKey Capital. So to ensure the data is being used effectively and transparently, we are introducing some sharing updatable models.
This framework allows us to host and train publicly available machine learning models in a decentralized, verifiable way.
So with some, we are making AI models more accessible and adaptable. For compact models, which are ideal for edge devices,
we can deploy smaller models directly on-chain to ensure rapid access and scalability.
For larger models, we utilize off-chain deployment through decentralized storage like BNB Greenfield,
allowing us to handle the heavy lifting without compromising on trust or transparency.
The system ensures that AI models are not only updatable, but also that everyone can verify the data and the communications behind them, creating a more accountable and decentralized AI ecosystem. Now I think let's dive into how the DNA architecture works to bring all of this to life.
At the beginning of the system, we have off-chain data collectors that use Xdata to collect data from platforms like Twitter.
And these collectors gather valuable real-time Twitter stress that fits into the system.
Next, we have the validation nodes. These nodes use local models on edge devices to validate
the data being collected. This ensures the data is accurate and trustworthy before it
is used for training AI agents. Then we have the computation nodes which provide the GPU power needed for data processing
and vector conversion. These nodes handle the heavy computational tasks, turning raw data into
something usable for machine learning models. And together, these components work in harmony to
create a scalable decentralized network while data flows seamlessly, is validated in real time
and is processed efficiently for AI agents to learn from.
This is the backbone of JIN, enabling us
to build the infrastructure for a fair and transparent
AI ecosystem.
So our AI ecosystem. So our AI data
the architecture as I mentioned before is xData. And xData seamlessly integrates with various data
sources allowing users to gather data with minimal manual intervention.
Using the golden feature, we leverage a distributed network to collect data, enhancing overall
efficiency and reducing the risk of server overload.
It doesn't stop there.
Xdata also employs advanced machine learning techniques to categorize and label all the collected data quickly and accurately,
ensure that the labels applied are correct and reliable.
With the themed coding, we can quickly gather labeled tweets based on user preference. preference on top of all that xdata also prioritizes user control and privacy by employing advanced
encryption and access controls will ensure that users retain full ownership of their data and
maintain privacy at all times this means data contributors can confidently participate in the
ecosystem and knowing their information is secure and they are in control.
So this data ensures that the data is gathered efficiently, labeled correctly, and remains
private and secure so everyone can benefit from the power of decentralized AI.
So we've prepared a demo to showcase how Xdata integrates data clashing, validation, and secure process in real time.
I think it will give you guys a clear picture of how Xdata works.
May I, like, show the...
...on DIN.
Let's get you set up in just a few simple steps.
Look for the Download section and choose the extension that matches your browser.
If you're using Edge, click the Edge Extension link.
For Chrome, select the Chrome option.
Next, head to the Chrome Web Store.
You can search for Xdata, or use the direct link from our website.
Click Add to Chrome, and confirm the installation.
Once it's installed, you're ready to go.
After installation, you'll notice the Xdata icon appearing on your Twitter page, right
in the top right corner.
Now, click on the Xdata icon.
Let's connect the wallet.
Once connected, you'll enter the Xdata.
Let's check out the mission section. connected, you'll enter the Xdata. Let's check out the mission section.
Here, you'll see different tasks.
For example, in this mission, the rule says that tweets need to contain specific keywords,
Xdata or AI.
Your job is to select your favorite tweet that includes these keywords, and we've made
it easier for you by highlighting it.
You can click the God golden button to reply.
After you reply, the tweet will be labeled with the keyword. The better the tweet fits the topic
and the more popular it is, the higher your score will be. This procedure automates the tweet
labeling process. Once you've completed a mission, you can track your progress and see how well you're
doing on the leaderboard. This feature not only allows you to gauge your achievements but also motivates you to improve
and climb higher in the rankings as you strive to outdo your peers. When you goad in a tweet,
it gets stored on-chain to protect data privacy and ownership. You can check out all your saved
tweets in your xData space, your personal secure tweet library. And that's
just the beginning. In the future, we plan to introduce labeled xDataSpaces, making it
even easier to organize and manage your data. And that's not all. As a reward for your contributions,
you'll also earn wafers, which can be redeemed for various benefits. By participating in these
fun missions, you're not just selecting tweets,
you're labeling tweets with the correct keywords and helping build better tweet data.
This data can be used for future AI agent training, making AI smarter and more effective.
And that's it.
With Xdata, you're exploring AI-powered data nodes, securing your data on-chain,
contributing to AI training and earning rewards, all while having fun. Ready to get started?
So in the demo, we show you exactly like how a tweet thread gets labeled, and behind it, there is a complete verification process to make the data useful for AI models.
So if you guys are interested about Xdata, we check it out at xdata.dinc.lol.
And thank you guys for your time and attention today.
I hope we can create a more open, felt, and trustworthy
AI ecosystem for everyone.
Thank you, Caitlin, for your insight, for sharing sharing and thank you again for joining us today.
Thank you, Tiana.
Okay, then finally, let's welcome Trevor, founder of Tagr.
Hi, Trevor.
Hello, good afternoon.
Yeah, hello, Tiana.
Yeah, good evening, good morning, wherever in the world you are.
It's Trevor, founder, Tiana. Yeah, good evening. Good morning, wherever in the world you are. It's Trevor, founder of Tagger. So just an introduction. Tagger is a decentralized AI data solution platform for data labeling, collection, management, and trading. So it's quite simple, like from Alaya AI and CIN, previously, you might have noticed how important data is to the current AI landscape. And I'll tell you, roughly over 98% of all data currently,
like previously we produced from the web two era,
so from the internet.com boom,
had already been consumed by current AI models.
So now we need new data coming up,
especially specialized data.
What is specialized data?
For example, medical data, agricultural data,
like for example, legal data,
all of these specialized data are what we currently lack.
And that's a big part of what TAGAR does.
So basically we allow anyone, so you can also do that,
anyone to come onto the site, even without medical knowledge.
You don't have to be a doctor, you don't have to be a nurse,
but you can label a CT scan of a lung nodule with simple
mouse click interaction so currently our main site is um down for um upgrades because we have a very
very large like a very mid-out date coming up very soon so i'll show you a brief video and then we'll head over to my testnet to show you the actual function.
Let me see. Thank you. Yeah, so basically we could see from the previous video there that there were a lot of medical data that we wouldn't otherwise be able to know where it is.
So how Tagger does this is that we have a pre-trained AI model that initially roughly
60 to 70% knows what a nodule looks like.
So since it already roughly knows what a nodule looks like. So since it already like roughly know what a nodule looks like,
it can at least have a rectangle around the nodule so that everyone knows where to label.
So once you see the target areas, just click around the irregular shape. And you can see that
the mask, the outline is very precisely labeled from the pre-trained AI model because it already
roughly knows the irregular shape,
which otherwise will be extremely difficult and very tedious to do a manual labeling path
in a traditional way without a pre-trained AI profile. So your input in Tagger, so as more and
more people label these difficult medical data, then our pre-trained AI model will increase in
its accuracy as well, because we feed
these data back into our model as well. And then the accuracy grows from 70% to 80% to 90%, all the
way up to 98.6%, which we have completed for a hospital. And the hospital currently uses the
model to do lung nodule detection, which would otherwise reduce pain in patient scan.
So that's basically the AI code pilot section.
However, I also want to introduce two features
in this major update, manual labeling and data review.
It'll be coming up very, very soon.
So currently I'm running it on testnet,
but let's head over there.
It's not that, but just note that it's not the mainnet,
it's testnet.
So it'll be live very soon.
Okay, so here I'm logged into my testnet.
That's the tag of mainnet.
So once you connect your wallet over here,
click task plaza and find manual data labeling.
I'll start with that.
So what manual data labeling is for is basically,
previously I said we had medical data,
we had agricultural data that rely on pre-trained AI models.
But what if we have a set of data
which we didn't train our pre-trained copilot tools on, right?
We need to label them from scratch.
So that's what manual data labeling does.
Labeling something from scratch
until a point
where we can use the data to train
a pre-trained copilot tool,
and then everything will become very, very fast.
So for example, a stop sign immediately popping up.
And right now you can click, for example, the polyline labeling tool.
And it basically just outlines the shape.
So that's very convenient for something like a polygon or like an octagon over here.
Okay, double click to finish.
Now I've successfully outlined the stop sign.
So I press confirm to submit this task.
Okay, what if I could be a bit slow here where I am?
Okay, so now we have the second picture coming up.
Same thing, I'll just do it very quick this time.
So in total you have 40 tasks to complete every day
per wallet.
So now, after I press confirm,
if I, for example, if I just did two tasks for this day,
I wanna finish my task and, you know, do it tomorrow.
So at the end of each working day, press finish session.
So we can see here, remember to upload completed tasks
at the end of each working day
to the task management page to the blockchain.
So be eligible for rewards.
So after finishing a session, the page will take me to the task, so to be eligible for rewards. So after finishing a session,
the page will take me to the task management page.
It shows basically each day I have been working.
So here, because I've been like on testnet quite very often,
testing things, I've been working on these different days.
So you can see that here's an upload button.
Once you press it, you can upload the task to the blockchain for rewards.
If you don't do that, you won't be eligible for rewards.
So please do upload because it uploads to a decentralized server for storage.
Okay, so that's basically the logic for manual data labeling.
You have to do it from scratch.
And then this is very valuable for our whole system
because it allows us to train a pre-trained AI co-pilot tool,
which would make everything so much more convenient,
like you saw in the first video.
And also like how we did it over the past few months
before the major updates.
Okay, now I wanna show you data review.
However, I need to switch accounts for that.
So, there with me for a minute. okay now i've connected to a data review account let me share
okay we also have a feature called um data review so essentially what that does is to operate all reviews and data cleaning within a peer-driven framework just like how you know bitcoin has
different miners and different people recording
on the ledger we work pretty much the same way in tagging z corp system so we're running the whole
corporation we're running the whole um data labeling essentially like company on a peer review
basis so people will be there to review whatever i've done previously so remember i was labeling
the stop sign there will be reviewers to review'll label the stop sign. There will be reviewers to review
whether the stop sign labeling has been correct.
To become a data reviewer,
you'd need to be placed within the top,
like select number of reviewers.
So like different number of data review
will be assigned
per window and we have like around one week
for every window to open.
So, you know, just be the highest stakers in the window
and we'll be fine.
Okay, now we see here, when we go into a data review task,
we see the stop sign being labeled by a red polygon.
So that's a yes, because it's very precise.
This one's quite precise as well.
You can see the red lines around it,
labeled by our dev in the dev environment.
Let me try to find a no in this case.
I think that's the one
i just labeled
okay this one's also yes don't worry like um if it doesn't fit this one's a perfect example. You know, we have an octagon stop sign,
but it's labeled with a square, so that's a no.
Okay, so once we, again, once we want to finish
the working day for data review, we finish the session,
and we go to the task management page,
and once again, don't forget to upload your task.
So otherwise, you won't be eligible for rewards.
That's basically the whole workflow, but I want to introduce a larger picture because
what I showed in the demo seemed quite simple, but actually, its inherent meaning is that
currently I've been in talk with a lot of car companies, in fact, including like Build
Your Dream and Huawei.
I've been in talk with them, and what they do every single year is that they spend millions and millions of um
rmb basically um on data protection because uh you know how in web 2 on the internet era right
your doctor flows on the internet very freely without your knowledge same works for them
they spend in billions of dollars to label for example um autopilot starter so car autopilot
driving system data and in that case they would want to protect it against other car companies
from stealing it that's right that's very normal however um you know when they have data labeling uh like 24 hours
it goes on for 24 hours and no employee can bring their phone and there's like full-on surveillance
cameras and also a lot of servers to protect everything however you know you it's still prone
to set because people can technically steal it illegally because example, when they sell their data to another company, or when someone in their system
decides to send it to somewhere else,
which happens quite very often behind the stages,
then it won't be traced and it can't be tracked.
Even if they do happen to find it in the future,
they need to go through a very lengthy legal process,
which they obviously wouldn't want to do. that makes for example build your dreams not being able to work
collaboratively collaboratively with a company like for example tesla or like xiaomi right so
for the same thing and what we do here is also we secure these data that you just saw on a data authentication protocol
where people's data will be stored, like the data I just labeled, will be stored in decentralized storage and they will be minted into something called a data passport. So that passport will be
handed to our enterprise client or whoever buys the data and in that case all of the data is traceable
or whoever buys the data.
And in that case, all of the data is traceable,
authenticate,
tradable and tradable.
Because currently in a web2 space,
data is not tradable.
It's very difficult to trade because for example,
if I charge $10 or $10 per set of data,
essentially you can buy it
and just pass it on to anyone out there.
And that's extremely difficult and like it's just
not feasible for a business to do that so they do not do data trading build your dream huawei and
like all of these car companies out there they trade they train data and they collect data
behind closed doors so one company might be spending one billion dollars on doing that
the other company might be spending $1 billion on doing that.
The other company might be spending another billion
on doing the same thing.
They can't work collaboratively.
But with the data authentication protocol
and with data trading,
they can each spend half the amount
and work collaboratively,
that anyone will steal it
because it's first secured on the blockchain,
second traceable.
So with the structure, we form a tradable and also a decorp economy.
So it's decentralized corporations.
So if we look at scale AI in the current Web2 landscape, so that's basically the largest Web2 data labeling company out there.
They have around 2,000 employees.
around 2,000 employees and as the 2,000 employees grows to for example 5,000, 10,000, 50,000,
then the edge managerial costs, so basically the operational and the managerial costs will
rise exponentially.
But that's not the case for Tag is SQL systems because you can see that all data labeling
and all review are done on a peer review basis and also assisted by AI.
So like essentially we can host, for example, 50,000
or even up to a million or 10 million different people working on the task at the same time
without actually raising the managerial
and operational costs that's needed to run a business.
Much like the Bitcoin network.
So many people are mining on the network every day,
but no inherent cost has been increased.
Here, we just increase the amount of number of people who come in
who's willing to contribute their time.
And you get tagged tokens.
Tagged tokens has already been generated,
has already been generated.
So you can buy it or either collect it
from doing our data labeling tasks.
And once you claim it from dashboard here,
income management, you can see I can request Portag,
but that's an assessment.
I request it, apply, then I can click claim
and it will go to your wallet instantly for yourself.
Or you can keep it as well.
It's a value token anyway.
Yeah.
And so that's basically your
salary or your compensation for your efforts we don't need essentially we don't need any funds
to participate just your efforts and you can claim it instantly on a c-corp scale on a
just imagine you're working in a corporation thank you so that's all for today
Thank you, that's all for today.
Thank you Trevor for the insight for sharing.
And I do have one question for you.
Do you have any tips for our developers who are looking to build AI on BNB chain?
Yes, I certainly do actually.
Because Taga has been, we've been doing this for a year.
And I've been looking into the market.
I've been a very, like, I've been very deep into the market.
And so basically everyone's been saying
that we need cashflow.
Like we need cashflow as an AI project in this round.
Like that's very obvious because no project
without a cashflow or real revenue can survive
in the longterm. That's basically just no project without a cash flow or real revenue can survive in the long
term. That's basically just no different to a meme. However, sometimes companies like Paget,
we've been extremely down to earth with our fundamentals. We find a lot of company and
hospitals to work with, but we realized that if you don't stay close to our user, if you don't
stay close to like these fun stuff, then people are just going to forget about you, right?
You can't survive in the long term.
So that's why we spent, like we actually distributed 21%, roughly 21% of our tokens through 4Me.
And, oh, by the way, it's also an innovative like token distribution mechanism where people do data labeling to collect that 21% share of tokens.
And we also developed our own mascot.
We wanted to integrate some aspects of the meme hype into our project.
Our project was initially way too much down to earth,
but then we realized, okay, we wanted some fun stuff, right?
We want to live in a market.
We want to generate a bit of trend in a short term
for us to achieve our long-term dream,
which is like all of the enterprise clients,
all of the AI tools that we have been developing.
So don't jump too far ahead.
Maybe I believe as an AI dev, always stay close to our user.
And also the reason why I personally, as a founder,
like I wanted Tagus to be token generators
is because we want people instantly being able
to receive rewards once they finish a task
and not be like, you know, just be promised to be like,
you know, for example, promised later on
where we TGE will give you how many different tokens
depending on your reward point.
That's just not feasible.
We want instant rewards. We want to stay close to users. So be very close to market. use how many different tokens depending on your reward point that's just not feasible we we want
instant rewards we want to stay close to users so be very close to market don't be like way too far
head always you know don't leave the market yeah yeah thank you trevor for the suggestion and yeah
and everyone to check out uh tak on X for the latest update.
And thank you all for joining today's workshop.
And we hope you have gained knowledge
to apply to your own project.
And see you next time.
Thank you, everyone.
