Projects
![]()
No projects found
Investors
New
DAO Treasuries
DATs
Lend / Borrow APRs
Chains
Categories
Spaces
One Trillion Agents Hackathon – Winners showcase
Recorded: March 21, 2025 Duration: 0:54:54Space Recording
Short Summary
The recent hackathon showcased groundbreaking projects in the blockchain space, including innovative autonomous agents, cross-chain integrations, and unique applications in military operations and subscription management. With over 2000 participants and a prize pool of $100,000, the event highlighted significant growth and trends in the intersection of AI and Web3.
Full Transcription
so ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប ប Hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, hey, the The Thank you. Hello everyone, welcome to our winner showcase.
We wrapped up one trillion agents, autonomous agents hackathon just a week ago.
It feels like that because we're still finishing up
the payouts to the winners.
So there is a lot of background work on our side,
but it feels like it was yesterday.
It was 2000 participants and 100,000 dollars in prizes.
It was one of the biggest hackathons in New York system
hosted by NearDev and Hackbox.
And yours truly, super grateful for everyone who invested their energy and time and thoughts and had space into this hackathon
and built all of those amazing projects.
This all happens thanks to our amazing sponsors.
Near Foundation, NearDev, Electric Capital, Proximity, Aurora, Near AI,
and more and more of those, including BD in HOT and Swarmchain.
And you can see all of them on 1t-agents.devpost.com.
So if you're interested what our hackers submitted or what kind of bounties there were,
if you missed the whole fun, please go ahead to DevPost, check the Hackathon gallery and you will find out what
what is cooking in in the autonomous agent space right now in the intersection of AI and Web3.
So I would like to welcome our sponsors. We have three of them today with us. We have Slava from Aurora. Hey Slava, did you like the hackathon?
Yes, yes, definitely.
Definitely.
Hey Nate, hey, what was your favorite impression of the hackathon?
For me, we've been doing this for six and a half years as an NFT
marketplace, now an AI engine.
We've hosted a lot of hackathons.
This one by far was probably some of the best submissions,
most creative implementations, and things
that we could actually play around with that we've had.
So it was surprising, actually, but very exciting.
Hey, we also have, I think we have Kendall from Proximity,
if I'm not mistaken, but he's not on screen right now.
Hey Kendall.
Yeah, I echo what Nate said.
Hey everyone. Yeah, I know I echo what Nate said. A lot of actually really cool usable agents.
I think we also saw some agents that were using Bitcoin, which is not a super
easy thing to do using some really cool tech that us and a lot of the sponsors have cooked
up. And then a lot of cross-chain agents as well, or chain agnostic agents, which is really
cool to see.
So what I just wanted to flag to everyone who's hosting hackathons out there, these
three sponsors received the most high quality submissions
because they invested a lot of time and effort into describing their bounties.
And I think for this hackathon, I think it was a little bit unusual because usually hackers are begging for creative freedom,
let us build whatever we want to build.
For this space, for AI space at Atmos Edges, it's a little bit different because this is new for everyone, even I think for everyone on this call.
This is relatively new.
We're still discovering what is there to build.
So providing solid bounty prompts, explaining to people what exactly do we expect them to
build is a key to success and the recipe for great submissions.
Talking about submissions tonight, today,
tonight for me, today for, I think,
the majority of people on this call,
we invited some winners to join us
and to showcase their project.
The showcase will be set by track.
We have three tracks.
And every winner who's present on this call,
we receive four minutes to do a quick demo
and to explain what their project is about.
And then if we have some time left in the end,
we will ask them some questions.
So let's jump into this.
We will start with a track that is called DeFi Agents.
I believe that we have two projects who won this track.
So if we can get them on the screen,
that would be just perfect.
Start with Balancer Assistant.
Hey, Manuel.
Hey, hi, Maria.
How are you?
Ready for your four minutes of fame?
Yeah, yeah.
All right, let's dive into this.
Sure, definitely.
Yeah, so basically we built Balancerancer assistant for our hackathon this submission.
Actually, we built three agents out of which DeFi track, we got this price for the balancer
assistant.
So this assistant is used to swap tokens across chains from your near valid bit a valid itself.
So mainly we are targeting EVM chains.
So starting like we targeted polygon and other EVM chains as well.
So as you can see in the screen, currently we can swap any tokens by giving the counter
addresses of the token.
But also we are also working on to enhance this user experience by giving the token symbol
itself.
So begin the scenes we are working on that right now. And secondly, what we have plans is currently
the execution happens based on the user commands.
But also we have plans to make this fully autonomous.
For example, we can make this agent to be fully autonomous
by making it work for maybe any trading strategies
or yield forming strategies across any chains out there.
So that is our second plan to make this fully autonomous.
So once we make that autonomous, I think then only we'll have,
we'll be using the full make use of this application on top of Bitte.
So that is one of the major future things that we are working on. And
so yeah, as you can see in the screen, we are right now signing the transaction on the
EVM chain. So once that is done, we'd be able to see the transaction as in the Explorer.
Currently, we did it in the polygon network. So also this could be integrated with other
EVM chains as well.
Currently, it is working in Polygon,
but it will be working multiple EBM and non-EBM chains as well.
So for this G5 track, I think this is our submission.
And we are very happy that it was recognized,
and we got this prize for this track.
Thank you for staying within the timing.
I believe, did you hack on sponsored volunteers?
I think you leveraged a bit, right?
Yeah, absolutely.
Okay. Maybe we can ask Nate for some comments and maybe some advice.
What would you advise to Manoel to keep successful building this project?
Maybe some features that he needs to implement?
Yeah, he did a few. He did Lido as well, which you can stake ETH and do that sort of stuff.
What would be neat is if, like right now you put in the contract IDs, if you could search
the tokens and then say, I'd like to swap this for that, then we'll get that thing verified
and pushed to the main store. I think that would be pretty neat to push that on the heads of all the bits.
Definitely. That's being already worked on in the backend.
So, yeah, we'll notify you once that is done.
Cool. And what we're working on is actually aggregating some of your tooling.
So not only will you have an agent where you can have your balancer,
but you'll have your Lido
and you can just merge the two together
and all of a sudden you have one agent
that can do these two different things
or three different things with facts.
So exciting.
Let's go.
Merge it.
Thank you, Nate.
Maybe we have some comments from Slava or from Kendall.
Just some wise words of advice. Yeah, we have Slava here.
Hi. Did you review your manual submissions back when we judged the hackathon?
Did you take a look at it?
Actually, no, because it wasn't suitable for the virtual chain bounty we were sponsoring.
So I kind of missed it back then,
but overall project looks great to me right now and like I wish you well. And like as with any
project, like I usually just say to people, just stay focused, like just do one thing well,
make your customers happy. And that's like a way to success.
Absolutely. Thank you. Thank you success. Absolutely. Thank you.
Thank you, Suleyman.
Thank you. Thank you.
Words of advice, questions from Kendall.
Would you like to add anything?
Looks very interesting.
I think, yeah, I echo, you know,
I think what Nate said about making it easy
to add support for new tokens.
And then, you know, I think a cool thing to do
is thinking about how you can pull in additional context
for users to know, OK, this is definitely the right.
These are the right contracts or the right addresses
I'm interacting with.
Uniswap has their list functionality.
Jupyter has their verified token functionality.
I think that that'll be important for a lot of agent
builders to just
help users gain trust in using the agents. But it looks very interesting.
Thank you, Kendall. Thank you, Manuel. All right, let's move on to the next winner in our DeFi
agents track. Bitcoin agent, do we have a presenter? Yep. Hi, welcome.
Hello.
Hi, welcome.
Hi, Maria.
How are you?
Hi, Maria. How are you?
Everything is great, and I think everything is going to be great for you, at least for
the next four minutes when we're going to play your demo in front of all of those people
following DevOps Live.
So let's go.
Yeah.
Thank you so much.
Let's get started.
Jim, everyone.
Great to be here.
We are team AlphaDivs, and we built Bitcoin agent
during one trillion agents hackathon.
It's an AI agent that uses near chain signatures
to interact with Bitcoin L1.
So what can this AI agent do?
This AI agent can create and send Bitcoin transitions
using your near account.
It can also provide information about your near account
and associated Bitcoin testnet address,
including their balances.
So the text tag which we use here was Bittag Agent, Nier chain signatures, and Bitcoin L1 testnet. By the way, we will soon be having this agent live for Bitcoin mainnet. We have a
roadmap and I'll talk with the Bittag team to get the agent verified. So a quick rundown of
how all of this is working, background is. The agent is running in the AI agent runtime.
With just your account using the signer MPC service,
you can get a signature for
Bitcoin chain and then you can just
broadcast the signature using our agent.
Let's just see a demo of this and let's
just see how all of this works.
I will just click on Launch Agent and here we are on the playground in the Bittay interface.
We will just ask our agent that what it can do,
and it should give us a list of operations that you can perform with this agent.
With this, you can create and send transactions
and you can also get blockchain information
like your balances and all.
So let's just query our balances to see how much Bitcoin
I have on this testnet address.
So it just shows me I have something like 0.04 Bitcoin
and this is my testnet address.
So let's just try to do something interesting
where we can just send a transition
on Bitcoin chain with this agent.
So I will try sending 10,000 sub-socies
to this testnet address
and it should create a new transaction,
which I can sign,
which will get me a signature
signed by the MPC service,
which I can then broadcast on Bitcoin network.
So here is the transaction and we can see that we have received a signature here from the MPC
service. So yeah, I think the demo is there. So yeah, so now the agent tells us to send this transaction to Bitcoin L1, right?
So we can just get this transaction hash
and ask our agent that, yeah,
just send this transaction to Bitcoin chain.
So we will just provide it the hash
and agent will do its thing in the background.
So, and you will receive a hash,
which is actually a transaction hash on Bitcoin network.
So we can just query this hash to confirm
that whether a transaction was sent or not.
So, as you saw, this was a quick demo
of some of the capabilities of Bitcoin agent.
And in this demo, you can see that
you don't need to directly interact with Bitcoin chain to perform any operation. Everything
is just handled through agent in natural language. And I would like to share a future vision
of our project. The goal is to basically have a full stack Bitcoin agent on here using chain
signatures that can basically tap into Bitcoin DeFi. And the very next step that we have in our roadmap, that's the next thing which we have in our
roadmap. And we're in talks with the team and your ecosystem to build and have this idea as soon as
we possibly can. So a lot of amazing stuff will be coming in next couple of weeks. And yeah,
it's exciting times and we are very excited to be building with the AI.
So that's all. Here are our handles. If anybody is interested in getting in touch,
so you can just give me, I just send a message and I'm always available to have a chat.
Thank you so much. Amazing pitch. I always like when people integrate their vision of the
future or at least the next several months into their
pitch. So Nate, would you like to give some advice? What do you think about roadmap for
Bitcoin agents? Yeah, definitely creative, seeing Bitcoin stuff also with the other things
that are popping up. Really, a mainnetnet transaction I think will be key to getting
it over to getting it verified and we'll promote as much as we can. So very unique.
Yeah, that's exactly the plan we have actually. Yeah, that's exactly what we are trying to
achieve. Yeah.
And what's cool with this is that you create that MPC account and not only does it work
with like bits of wallet, but any other MPC related wallets will also work automatically out of
the box. So you're not just stuck with, with bits of wallet or any of the other wallets.
It's MPC related. So very neat.
Yeah. Thank you so much. Thank you so much. Yeah.
Slava, Kendall, any wise words for our winner? Kendall.
Yeah, super cool. We're very excited to see more Bitcoin functionality.
I think that's really taking advantage of the tech in a unique way.
I like the roadmap of focusing on some of the things you can do on Bitcoin,
like, you know, particularly the air drops, I think is unique.
You know, it's not just interacting with runes, but being able to distribute runes in new ways
is something that we haven't really seen a whole lot of.
Super cool.
I think focusing on onboarding, obviously it's super easy to onboard, which is great.
I think thinking about how you can make it easy for people to onboard their friends.
So one idea there is you can actually make it so that people can send a link that their
friends can use that will help them claim a little bit of Bitcoin once they sign up
using Bittowallet or any of the other MPC wallets.
So happy to you know help you guys a bit more with that if that's of interest we have some code for that.
But yeah I think just you know figuring out how you can get as many users as possible using this
because it really will be one of the easiest ways to interact with Bitcoin which is a really cool
I think onboarding experience for net new people.
Yeah thank you so much and very very interesting idea of like sharing a link
where you can just claim Bitcoin directly
using your video wallet.
Yeah, thank you so much, Kendall.
And a comment from Slava, from Aurora.
Yeah, thank you for your presentation and demo here.
From my side, I just wanted to mention also
that here at Aurora, we have the virtual chains,
the EVMs, which are settled on the new protocol.
And they are also like they can use the chain signatures.
And it means that they like you can reuse the products, like different apps on the EVM on Bitcoin.
For example, I've done a small proof of concept a few months ago with Gnosis Safe, basically
allowing the multi-signature wallets being used in Bitcoin.
And maybe that also could be an interesting direction for you to research.
You can go to the documentation, which we have
on Aurora.doc, Aurora.dev and take a look at the chain obstructions there. There is that proof of
concept there and maybe it could be interesting to incorporate some Ethereum-like product projects,
which could be used as like a business logic, but for the Bitcoin network for the future
be used as a business logic, but for the Bitcoin network for the future development cycles
of your product.
Definitely.
Thank you so much.
We'll definitely have a look at it and we'll connect with you to listen more about it.
Yeah.
All right.
Thank you so much.
And just a quick note for all of the other projects out there, whether you won the hack
or you did not.
We have a chat on Telegram,
which is called near-dev at near-dev.
So if you post your project description,
your pitch there with the hashtag roast me,
you have a good chance of receiving some feedback
from our judges and dev roles.
So this stream is not the only opportunity to do so.
Now we're moving on to the next track, social media agents.
And I think Project Comchain is the next presenter here.
Hi, Silver.
How are you feeling?
Yeah, I'm good.
How are you?
I'm doing well.
So let's get ready to your four minutes of fame.
So you'll be showing the video.
Okay, okay, okay.
Perfect, so my entry was Comchain.
So it's for actually, it's a serious business.
It's actually for military operations.
So what we see right now is that the frontline tasks
are not completed by the soldiers.
We have lack of verification of conditions. There's also payout
issues because in military you have rewards, you have bounties, but for tasks to be completed,
but there are issues like corruption and other payout issues. There's actually a lot of
science done around this and blockchain shows potential. So we're introducing something called
like decentralized field operations,
where you can actually do it in secure and private way,
use a near AI and also encrypted data
to actually get much more things done.
So we also have token based accounting
and rewards in the system.
Okay, so let's see how it works.
So first of all, you log in
and the idea is that
you can assign tasks to frontline soldiers, and then the soldiers can then complete the tasks
and then earn bounties. So how does it work? So here on the left bottom side, you can see the map of ongoing activities.
Here on the left, you see a list of things or tasks that can be done.
For example, this one is conduct drone hit.
So here soldiers can upload evidence of successful drone hits, for example, and then the AI will
verify it and then they get a payout.
So here is just an example where I upload a photo.
So that's a task to repair water stations.
So I upload a photo, it gets encrypted,
put on storage, then a smart contract that will be hit,
the task, the completion will be stored there.
That smart contract will trigger near AI
that will verify the photo and then some time goes by and
then it will actually say okay the photo contained the evidence I accept it it's fine so now the
commander can now reward it right so they can actually issue payouts to soldiers so how does
it work that's the schema of all of those tasks.
So you see the task ID and then all of the like the name of the task, but importantly,
you see the instructions. So you have the instructions for the soldier, but you also
have AI instructions. So these are for very fine evidence, whether it's a video or photo or whatever
it is. So this is the data that gets stored on blockchain. So you have the result of the verification, the evidence,
and so on.
Now, here you see the Near AI as it queries blockchain,
and then it groups the evidence.
And then it will describe the image
using this feed-free vision model from Near AI.
And that then uses it to understand what's on the image.
So the other applications that I can see is disaster recovery operations, humanitarian
aid, social rescue missions, crowdsourced data and so on.
So it's not only about military really, it can actually help in other fields like massive
field operations and to help to complete them and their tokens and then everything is also secret and
decentralized. So that's the entry. That's it. Thank you, Silver. Talk about real life use cases
for blockchain. Dear judges, do you have any thoughts? This is quite unusual, I think,
for us using blockchain formulary. Nate, do you have anything to say about this project, Comchain?
Too deep into it, so I can't give a bunch of information, but yeah, keep on going, push,
push, push.
Yeah, so I saw that there's a lot of research on how blockchain could help in military,
but I don't know any actual projects, so I thought I should give it a try to see,
especially with secret AI, like near AI is, I think it's a very interesting proposal for these kinds of things.
Might be, I think it's a little bit of our prey, I think it's great to judge that, but maybe Slava has an opinion.
Yeah, it's hard for me to comment on the military applications,
but overall it looks interesting just in terms of completing and improving the tasks.
Thank you. Kendall, do you have an opinion on this?
What do you think?
At least on the privacy side, yeah,
I mean, fortunately that's an area
that we are very focused on.
I mean, there's a few different options there,
but like I think the core idea is you can actually run
a lot of the logic within the trusted execution
environment cloud that Nira is building,
or even bring your own.
So like we have this kind of protocol called shade agents
where basically you can have like a network
of independent Ts that are running code
and they actually can prove on chain
they're running the right code in the T.
And that way you can do kind of like end to end encryption
where like the user can basically encrypt the data
using a key that only exists
within the trusted
execution environment.
It can then be decrypted, operated on, and then re-encrypted.
So even the T operator doesn't know it's running.
So yeah, very focused on privacy.
We're also thinking about how to keep addresses private as well of agents.
It's just a little more complicated because the way we do key management is using this
like decentralized NPCs.
So there is a little bit of leakage of the keys,
but if you use some of the pool-based private solutions,
there are some options.
So yeah, I think there's a lot of interesting applications,
definitely happy to kind of chat more
about how to enable that.
All right, thank you, Sivarsap.
And I hope you will keep building
because I remember you from one of the previous
hackathons. I think you participated in every big hackathon near hosts. Let's see what comes up
as your next project. So we're moving on to our next track, last but not least, absurd agents,
everything that cannot accommodate DeFi agents and social media agents. We have a project that is called PING
Subscription Shade Agent and we have Jay to present it. Hey Jay, what is this about,
the PING Subscription Shade Agent? What does it do? Hey Maria, thanks for having us. I'm super
stoked to be here. I think we were both, like me and Elliot, were both really excited about the
hackathon. So yeah, PING is essentially taking a new approach to automate and reoccur in payments,
our subscriptions on Nia without taking the approach of pre-funding the subscription for
a specified period.
And we did this by using the Shared Agent template.
So in the demo, you'll see from the merchant perspective, the dashboard shows all of the
customers subscribed to the different tiers that the merchant offers, highlighting the
next expected bill and date and whether the status is active, passed to you or
cancelled. And then from the customer perspective, you can see all of the
available plans that the merchant offers and select from a series of different
payment periods. You can then go to checkout, connect your new wallet and
fill out all of the relevant details details as well as selecting a maximum subscription duration and set if you want to sit six months as
your maximum duration for subscription you can do. The reason for this approach is that in the
current crypto subscription models customers typically receive email reminders at the end
of each subscription period or prompting them to manually renew,
or pre-approving for a saturation upfront,
which is inconvenient for managing subscriptions.
So as you can see,
we're going through completing the subscription,
and then you'll be able to see the active subscription
in the merchant dashboard,
which we're just putting to now.
So then instead of requiring manual payments at the end of each subscription period,
Ping's Shade Agent will handle the renewals automatically to ensure there is compliance with the contractual subscription terms
and require no further action from the customer other than pausing or canceling the subscription.
And then as the renewal date approaches, the shade agent will autonomously process the payment
on behalf of the customer.
So we're currently looking to improve the demo
that you're seeing here from the hackathon,
fix some bugs and build on the current logic.
But what we do do currently is we provide an SDK
for merchants to integrate their own sites
for users to create subscriptions.
Additionally, we're gonna provide an SDK
to create user dashboards, giving them full control
to cancel or suspend the subscriptions
without relying on the merchant and respect the decision.
The solution we see is completely generic,
allowing support for many merchants to utilize it
and simplify it for users to manage all of the subscriptions.
And as you've seen in the demo now,
there's just a little bit of a technical overview
of the reoccurring payments.
I'll just do a quick two second through that.
So for this, for five steps,
there's the subscription creation
where the user approves the initial transaction,
creating the function call access key pair,
which allows the contract to initiate the future payments.
That public key is stored in the custom contract
while the private key is stored in the trusted execution
environment private storage.
The access key may have an allowance
which would limit the amount of money
able to pay for subscriptions, which
enables an automatic expiry, which doesn't really
exist in current solutions in Web2.
Then the remaining steps are the shared agent monitors
and triggers the payment operating in the trusted execution environment. So it checks periodically for the upcoming renewals,
it verifies if the user has sufficient funds and then triggers the transaction using the key
pair on the user's behalf and then updates the subscription status in the smart contract.
Marginate is renewed and then sets the next billing date, kind of keeping that recurrent payments going.
And then in terms of cancellation of management, users and providers
of subscriptions can cancel the subscriptions via the smart contract.
And no further payments would be deducted from the user until the end,
and they would have access until the end of the current subscription cycle.
So yeah, that's kind of a quick run through of the submission. We're really looking forward
to kind of building this out further for both reoccurring payments, but also
general single use payments for merchants, users, and communities. Thanks.
Thank you, Jay. And I believe that shape agents were initially shaped by proximity.
So maybe Kendall, you have some advice for Jay?
Yeah, it's definitely an interesting way to handle subscriptions, which are obviously always a kind of complicated thing for
for crypto based payments. Yeah, I mean, I think like the key for that, especially for appealing to the Web2 audience, is thinking
about on and off ramps and how to just abstract away as much of the crypto side of things
as possible and really highlight what some of the advantages are, which I think you guys
are doing.
So yeah, definitely interesting.
Mainly just, I think, figuring out who the early adopters are going to be is key.
But yeah, it's definitely a fascinating solution.
Very, very cool.
Awesome, thanks Kendall.
Thank you Kendall.
Nate, Slava, do you have any comments for Jay?
Yeah, pretty neat.
I mean, it uses Nier pretty heavily, the allowance.
So it would be kind of like a debit card
where you fund, you create an agent
and then you send funds to it and it
automatically just spits out to all your connected payments. And then is that how that works?
Yeah, so there's kind of a few different like further use cases that I see available from
it. Saving accounts being one and then like automated direct debits essentially kind of just a proven
that I want to send this person five pounds per month or five dollars or whatever currency you
want to you want to go with and then that would happen until you want to do it so I think that's
kind of good for reoccurring payments to friends or family as well kind of could be used as a cool
onboarding tool and then peer-to-peer commerce as well I think is also awesome for that.
And this thing is just running in the TEE, just executing on the per month basis on whatever you do.
How close is it to, is it, could it be main net right now or is it still still in testing mode?
So myself and Elliot are definitely going to do some tweaks and try and get things done as soon as we possibly can.
There's definitely a few bugs that we need to iron out and just make sure it's all wrapped out really nicely
for an end user experience.
But yeah, kind of excited to start shipping that soon.
Yeah, good idea.
And it would be your first customer, I guess,
within the near ecosystem, because I'm assuming
those would be the.
Yeah, I think there's a lot of good options.
I think I see it as opening up a nice revenue
model for a lot of the AI teams if anyone
wanted to start doing subscription usage
for any of their models or anything like that.
And then there's community initiatives
with monthly subscriptions for different merch drops
and things like that from the likes of Black Dragon
Community, that sort of thing.
I think it would be a nice first roll out.
Cool.
Thank you, Nate. Thank you, Jay.
Slava, do you have any comments, any advice for Jay?
I see just really good and prominent,
how to say, usefulness of this project overall.
And it's really nice, pretty practical product.
I see a big perspective in using it in terms of like subscription
service for any DAP on Nier and as well Aurora.
It was possibly could also be done for like Utah chains via
cross-contract call communications, et etc. So yeah, it looks good. I like it.
Thank you, Slava. Yeah, we're definitely looking to kind of use near intents to potentially offer it
like cross-chain to pay with any currency for subscriptions too. So definitely something we're looking at.
Thank you so much, Jay. And we're moving on to our next project, AI Agents in Need of Escrow, Volodymyr.
Hi, guys.
Are you with us?
I am.
So let's go into your demo.
Sure.
Happy to drive that.
All right.
My project is named AI Agents for Need of Escrow Platform.
So during this hackathon, I have implemented two AI agents.
First agent is Compliance Validation Agent,
and second is Disput Resolution Agent.
They are very needed for my Need of Askrow project.
I have developed API on top of AI agents,
so the communication with them is more convenient.
And also this API has been integrated
in this Need of Askrow project.
A few words about this Need of Askrow project.
This is a web-based platform designed
for secure online collaboration.
It is powered by near-protocol blockchain,
and it functions as an Askrow service, holding deposits
and releasing them once specified conditions are met.
And AI agents are needed here in two places.
First place is dispute resolution process.
Dispute resolution agent helps the admin
with resolving the dispute.
It helps with providing its own dispute resolution opinion
and further clarifications.
And another place is task creation process.
This is where compliance validator agent helps
with natural language processing
or the task name and task description.
So the task content is validated
according to platform compliance rules.
And now we will jump into the demo.
First, we will see how this resolution process looks like.
We can see a task name and task description here.
We can also see the task is under dispute resolution
process, and there are comments from task owner and task
contractor at the bottom of this page.
And the goal here is to pick up the right resolution
percentage.
The admin, of course, can do that on his own
or use the help of AI agent.
AI agent will take the task name, task description,
dispute resolution comments, and will provide its opinion
about resolution percentage.
Let's see what this agent responds with.
It thinks that 70% is a reasonable resolution.
That's fine.
And also admin can talk to this AI agent
to clarify this resolution further.
And the next place where AI agent is needed here
is task creation process.
A natural language processing of task name and task description
blocks creation of the tasks, which
violate the platform compliance rules.
So here I am trying to create a task which involves
some kind of harm to animals.
Let's see what AI agent responds with.
And we see that it actually blocks this task creation. So this is it from
my side. Thanks a lot for having me today. And thanks a lot for such a great opportunity
to try developing something with near AI agents.
Thank you, Volodymyr. I'm just reading your page in the app post about this project, and there is a prompt that asks how we built it, and you said easily.
Was that your experience with NearAI?
The thing is that I am not using some kind of cross-chain communication or something like that.
These are simple prompts here, but the main thing I am using from this AI agent is that they are isolated.
So my tasks and task description will not
spread into, I don't know, different areas.
And maybe other people can see that tasks, this name
and description could be under NDA.
So this is quite important here.
And yeah, but other than that, this is mostly web 2 stack
rather than web 3.
Thank you for the answer.
Nate, Kendall, Slava, do you have any comments or advice
for Vladimir?
We will start with Nate.
Oh, I don't.
I haven't dove into this one enough to really give
too much input.
Sorry.
All right, then maybe Kendall or Slava have some input. If you do. Yeah, we have Kendall.
Yeah, I mean, I think this idea is really powerful. Kind of using AI to mediate disputes. Yeah, we were talking to, yeah, I mean, or like that, I know, like that, when my team who obviously plays a
big role in this broadly in the crypto space is very interested in the concept. I think the big thing
to figure out is well, two, twofold. One is like helping users gain trust in, in, in models to
mediate disputes. I think, yeah, starting, starting with like kind of more minor disputes is probably
a good way to like things that were speed and just an absolute decision or more important than like
absolute, absolute correctness, I think is like a good potential approach. So like building up a
confidence there and then obviously get, you know, making the model, giving the models enough context
to actually be truly correct. Because I do think there's yeah, there's an interesting future where
many, many disputes will be mediated by AI probably they'll always be something that's like kicked up
to humans if it's like super, you know, like vague and like kind of more of a
discussion of like broader morals. But yeah, no, super powerful. I think yeah, there's, there's,
I mean, I'd love to see this in production for some things like very, very quickly because it's,
I think like it's clearly where, you know, where things are going.
Thanks for the feedback. Thank you, Kendall. Slava, do you have a couple of words for Volodymyr?
I can kind of continue the thought Kendall was saying that it could be thinking about support, like maybe, you know, how good
the support ticket was closed and maybe given some points instead of real money to like gamify,
I don't know, Discord support server for Nier or Aurora, something like that. It's just like on
the level of idea, but could sound really cool by using the agent and these models.
idea, but could sound really cool by using that agent and these models.
Yep.
That's great idea actually about support.
Yep.
I'll take it into attention.
Thanks, John.
Thank you everyone.
At this point, we will wrap up the project showcase and do a micro
mini panel with our judges, because everyone who is present on this call,
Nate, Kendall and Slava, all of them have judged submissions for their
own bounties. So I would like to hear your opinion. What was built? What did you like?
What did you probably not like very much? And some advice for the hackers who won this
hackathon, maybe didn't win, but they want to keep building, how to turn the hackathon
project into an actual startup. So let's start with Nate. Share some of your interesting outcomes from this hackathon.
How did you judge your bounty, and what interesting project
did you encounter?
Our biggest aha moment is that we
didn't help a lot of the devs to do what they did.
I think we had 18 of the top 40 submissions and probably 10
good projects that we were able to mess around with
and actually test ourselves.
And so, I mean, it was just really nice to see people figuring out the system
without needing to do handholding.
So that's a pretty big signal to say, okay, we're on the right track,
that the onboarding process is incredibly painful,
that people can build agents.
Another neat thing is people took it way further than we thought.
We had just announced this embeddable chat system, probably, I don't know,
like a week after, right before it was about to come.
And two of the projects went that step, like Coolify embedded near
intense into their own web page.
Seeing those things kind of taking
it the extra step as we're building and shipping things they just take that new
stuff and throw it in and it's it's it's neat to have folks messing around with
with all the things that we're we're shipping so and just such a wide variety
of submissions literally Bitcoin someone created satslinger, which was the, if you fund a Bitcoin account,
then it will tweet to people as long as you tag them with certain hashtags.
And it will actually tweet to them as long as there's Bitcoin into an
into an MPC account that they can claim their tokens.
MPC account that they can claim their tokens.
Fascinating.
Fascinating.
And then again, seeing the, what is it?
The bet facts, Owen built this thing
where he could do this whole betting system
and it really shows the power of no longer needing forms,
having to input forms, right?
You're having conversations with applications now.
It's not, here's a input for this, an input for that, an input for that.
It's more of, I don't know how to bet,
show me the leaderboard, what did the last person bet?
Okay, copy that exact same bet that that last person did
and then put 500 bucks on that bet.
We're getting to this new point where
we're gonna start seeing there's no need for just how many crazy things we
can do when you just open up what AI can do for us.
All right.
So if you have to give just one advice for people who want to keep building, what should
they look into?
What should they do to become a startup?
Becoming a startup is the most brutal thing a human can probably go through.
One of the few.
There's a lot of other brutal things,
but when you start a company, at the end of it,
it probably won't be your same idea.
You just crank, you believe in the long-term vision
and you just go with the flow.
And it's, you just, grit is probably
the most important element of making a startup a reality.
If you don't have grit, it's not going to go anywhere.
There's no magic.
No one's going to help you.
You just have to believe in yourself and keep cranking.
So if you, if you have that, then go, go, go push.
Thank you so much.
And obviously taking this from an actual company owner, I think it's a very, very valuable advice. So we are moving on to Kendall. Any interesting submissions for your bounty? What did you like the most? Which one?
Yeah, well, it was very cool. I mean, I think we're just super excited about expanding what people can actually do with Bitcoin.
I mean, Bitcoin is the most important crypto asset and I think it's extremely underutilized.
So it was really cool to see, you know, yeah, teams like, you know, like the Bitcoin agent, like the Satslinger and there were a couple others as well, which is really exciting. And I think, yeah, teams should really think about, like,
what can we enable people to do with their Bitcoin
now that we can make it much easier to use and also cheaper
to use, right?
What's cool about the MPC tech is
that you can actually send out, you
can transfer the ownership of a lot of Bitcoin
without having to touch the Bitcoin L1 in a way that's
potentially a bit more programmable and interesting
than some of the other solutions out there.
So I definitely recommend teams look more into that.
Yeah, I think as far as what to do next, something.
So we released the Shade Agent concept
pretty late into the hackathon.
So it was really cool to see a few teams pick that up,
like PingPay.
But obviously, it did come pretty late.
Just to give a very short recap on what it is
and why it's interesting, it really
is basically about just making the entirety of your agent
code base both fully auditable, verifiable, and then also just
enabling any number of participants
to actually run your agent in a verifiable way, which is, I think,
one of the key aspects to making agents unstoppable.
I think what the use cases and why it's interesting
is you can think about ways that you can actually
embed an agent very directly and deeply
in kind of a crypto protocol.
Like, imagine if you took Aave and it
was agents who were actually making all the decisions of Aave instead of people. Whether or not that's a good idea today is
a matter for discussion, but that's clearly where things are going. And I think shade
agents enable that in a pretty interesting way. So I advise people to kind of think about
how that could work out. As far as making companies a business, I mean, I think Nate
gave some incredibly sage advice there. Obviously, you know, coming from a place of serious wisdom.
So I encourage everyone to take what he said to heart.
And then, you know, just to extend on that a little bit on the kind of like the product
side is like you solve problems.
I mean, I think there's a lot of really cool agent demos out there.
I think some of the groups here have built have actually started to really solve problems.
But yeah, make sure your agents need to be useful and they need to actually do
things.
I think that's like when you're going to see really valuable companies built on
top of this.
It's like agents that actually do things, solve real world problems in ways that
were impossible to do before.
So encourage everyone to think about that.
Thank you so much, Kendall.
Sage advice right there.
So Slava, what did you like the most about the submissions for Aurora?
That's, that's interesting question.
Uh, probably like I have my favorites.
I have my favorites here and actually like my favorite submission hasn't
won, hasn't won a prize.
Unfortunately.
I mean, the first three prices, even the Aurora price. hasn't won a prize, unfortunately.
I mean, the first three prizes.
Even the Aurora prize or the major prize.
We got them like not on the first three places
because the idea was like super, super idea,
really good one, but unfortunately the team
hasn't developed enough like smart contracts and haven't done enough
solidity development.
And because of that, we just couldn't give them anything in terms of the main prizes.
But we have given them some money to prosper that idea further.
We just had a call with that girl who was,
which was working on that idea.
And basically what she had there was like
a competitive battleground for AI agents.
So basically every AI agent, it's like a character
with different like RPG like characteristics.
I mean, like agility, maybe strengths, something like that,
maybe some magic powers, maybe some spells. So there is like a set of actions that agent can do.
There are some actually like how it's called like fractions or like clans of AI agents,
which kind of fighting in between. And you kind of is an observer,
observing the battlefield where these agents are fighting
and you can do that.
You can put some money to improve some clients.
So it could be really interesting mechanics
in terms of gaming experience and getting involved
or just having a content being generated by a
form of agent fighting between themselves. So the idea was really good. And we hope that
in the future, you'll see something from them. But unfortunately, she probably just hasn't
enough time to work on the project during the hackathon.
And that's why it wasn't like evolved into the final project.
Talking about our...
I just have a question because I don't need to interrupt, but maybe Aurora will have a hackathon that this particular participant could submit, if she has enough time.
Yeah, of course.
Which ones? Where should we follow your announcements?
What?
Where should we follow your announcements?
Or maybe you have something particular in mind.
On Twitter. The best way to follow us is just go to Twitter, Aurora is Near, and follow us there.
We also have the Telegram channel, and you can follow up on the Discord, but probably the main way to follow us is x.com, Aurora is Near.
Alright, so any advice on how to actually turn the Hackathon submissions into a proper company?
advice on how to actually turn the hackathon submissions into a proper company.
Like, I don't know about the practical side as good as previous judges.
But in terms of like just pure reasoning, I believe that like you just need to be
focused and also ready maybe to pivot and maybe change
something in your product to be more suitable to the market conditions,
you know, because it's like waves sometimes like something about the
IS comments and sometimes something about the device.
or DeFi, sometimes you need to combine.
Sometimes we need to combine.
So it's like you need to be flexible
and solving the real problems of real people.
And of course, like finding support.
And by support, I mean investments,
like maybe incubation programs,
which can help your project grow
and succeed in the future.
Thank you, Slava. Solid advice.
And I think of all three judges presenting,
Nate was the only one company owner.
He pretty much said that this is a brutal business and don't.
I would encourage you to actually try again
and maybe join one of the upcoming hackathons,
maybe Proximity will organize something, maybe Aurora will organize something, but we are already organizing something.
There is a hackathon ongoing, ZcashX, near intense hackathon, it's also dedicated to autonomous agents, we already have 76 participants, competing for a 70k price pool. We already have our first submission. So if you have a
Taikai account, you can actually log in, register for the hackathon, take a look
at what people are building. It's pretty interesting. So if you missed this
opportunity, 1 trillion autonomous agents hackathon, or maybe you have some fresh
ideas that you would like to try.
Join this hackathon, we would be super thrilled and excited to see more of your submissions,
because it's always super interesting to be on the edge of exploration in tech.
We know that AI is the next frontier, so let's keep building it together.
And thank you so much for joining us today,
tonight, this morning.
Nate wants to say one last thing.
And so we're gonna let him, I guess.
Also, I wanna say thanks to the Vax folks.
They're a DeFi liquidity pool system up and coming on near.
They built a really neat engine too.
So I wanted to give them props as well.
A really interesting agent that will be getting verified shortly.
And they had a nice bounty as well.
So good job, Vax.
Let's get it going.
Thanks.
It's actually, yeah, it's an odd one because this was the sponsor that
joined much later than the main batch.
And then we saw some amazing submissions.
So thank you for everyone who supported one, truly. And justathon. Thank you for everyone who keeps building in the new ecosystem.
And thank you to the winners who joined us and raced it and decided to present their projects
live. And I really hope that you receive some valuable feedback that you will take home and
keep building. So that was Hackvogue. Thank you you so much for hacking with us and until next time!
