NEAR Protocol Town Hall: January 2026

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:06

   GM GM everyone and a happy new year to you all. I am Peter from New Week and super delighted to be back with the very first town hall of the year. 2025 was a remarkable year and seriously everything that happened in 2025 I believe is only a fraction of what is going to happen in 2026 for the NIA protocol. We've seen great traction and great product launches. Now it's all about getting

2. 0:41

   that adoption and users and today on a town hall as always very delighted to have Ilia Pulisukin. Then we're going to have some AI stuff from Cameron Dennis. We're going to hear more from the core protocol team from Bowen and Nean. And lastly, we're going to top it off with a little bit from the events team on some of the many great events that are taking uh place this year. As always, I'm going to bring you to the stage.

3. 1:10

   Ilia, how are we doing? Hello. Hello. Happy New Year, everyone. Great to be here. All right, Ilia, please let me know when to change the slides and it's all yours. All right, let's go. Let's go to the next slide. Yeah. So as this is the first town hall of the year, I really want to kind of kick it off and discuss how we make this year of near. Let's go to the next slide.

4. 1:42

   So just kind of looking back at 2025, right? We had obviously Near Protocol uh publishing on Twitter, publishing the review with a lot of stats, but kind of key highlights really are that near protocol um has truly proven kind of that it's infrastructure that can handle both load and um operate at low latency. We had 1 million TPS been shown really showcasing that sharding works and can scale. uh with more shards you have more

5. 2:15

   throughput. If you haven't read the near one blog post, it's really great kind of showcasing that uh you can we can continue scaling the network. 1 million TPS is really more than you need, right? It's Visa, Mastercard, Stripe, and a few other uh major payment processors combined. uh and uh multiplied by 10 that's roughly what millionth TPS means and 600 milliseconds right is really kind of getting close to uh web two to

6. 2:49

   speed um for every processing and everything obviously there's continuous work on optimizing that this year but all in all mirror protocol is there it's proven infrastructure it it hasn't went down and it continues scaling continues getting faster Now near intense uh really went from kind of an idea in 2024. We announced it at redacted and through last year really achieved product market fit. Right? We now see

7. 3:21

   people implementing it and integrating it. Uh we just saw today MERT vibe coded it with claude integrating into Orb. So really exciting to see this this product that kind of uh spreading through the whole web ecosystem already and it's a truly rare feat threat in crypto like most of the especially infrastructure is just uh stays there and it's generating fees for near uh we're going to have more updates on kind of exactly how it's going to work. Um it's you know it's not

8. 3:53

   as easy because there's a lot of partners and we want to make sure that it's fair for everyone but uh It's definitely going to be a massive engine of uh fees for near and near token. Now, nearai uh launched its cloud and private chat in December. It's still early days obviously. It's still been only five weeks, I think. Um but, you know, it's already has a number of customers and and more in the pipeline really a lot of feedback kind of uh and

9. 4:24

   uh next features are being built right now. uh all of it is open source so you can kind of continue tracking. So really exciting uh developments across kind of core pillars of this vision that we've outlined you know a few years ago. Now if we go to the next slide what I see is happening this year is really bringing this uh kind of pillars together right I think um kind of right

10. 4:55

    now people see near intense as you know hey it has product market fit why don't you focus on that solely uh there's other people who are like hey new AI is awesome you know privacy verifiability is extremely important there's a lot of enterprise need for this um and so kind of there is a there's a ton of uh interest on both of the sides but the reality is it's about how to bring them both together and so uh really this idea of agentic commerce of where AI is becoming both the

11. 5:30

    purchaser and the merchant um kind of the operator the executor that's where we see the convergence of it and I think this through year we'll see more and more of this uh coming together. Next slide. Now why is this happening right now?

12. 5:49

    Well, the traditional kind of web to stack is starting to crumble, right? I mean, I'm sure you all seen cloud code and cloud co-work uh the current tools are not designed like Google, Amazon, the advertising model, the websites, the documentation like all of this we see traffic dropping, the advertisement uh revenue is going to start dropping.

13. 6:16

    It's all been optimized for user attention, right? grabbing it, monetizing it. It's not optimized for agents executing actions and really getting things done. And obviously, you know, some of the companies have long prioritized now the revenue over experience on the business side and business commerce, it's even worse, right? If you're imagining when you need, you know, to do, I don't know, textile

14. 6:40

    production and you need some chemicals, right? Right now it's a pretty manual process still because there's no really a market where you can do that. There's a whole process of invoicing, billing, the legal work and all of that goes into the cost of items that being produced.

15. 6:55

    Um there's you know cost of capital that's stuck in flight uh while you know the product is being delivered in while invoices are being done. So you know on the traditional e-commerce side which always been evolving faster we see just shift of user interface right I mean something I've talked about AI is going to be the user interface right even just in last few weeks this kind of transition accelerating and on the

16. 7:19

    business side there is kind of now we have all this intelligent tools but the kind of core of the business is really still slow manual and kind of uh really hard to negotiate and so there's like this immense opportunity to really bring those pieces together. Next slide.

17. 7:40

    And so there's a better way, right? This is where, you know, we bring what crypto done to, you know, financial markets, really turning them 247 instant settlement, uh, self-custodial, compos compos composable, and really bringing this to the rest of commerce. everything from e-commerce to trade finance to you know traditional kind of factory supply chain to uh really kind of larger and smaller uh projects and PL plans. And so

18. 8:11

    this is where intents which are really about declaring outcome and finding a party or parties to work with and agents which are going to be those parties that actually can navigate negotiate and execute. uh coming together bringing us this unified commerce. Next slide.

19. 8:32

    So AI agents are becoming economic agents and they need their you know interface and they need their way to execute this and so intent is that way for them to action things right where they describe the outcome they want and they can find a counterparty to do this.

20. 8:48

    Um, I mean, as simple as, hey, you know, need to buy a pizza. I want to get it delivered by 7:00 p.m. to, you know, need to have some, uh, you know, tons of steel delivered to this warehouse, uh, at this quality at, you know, within this parameters. And so, this agents are not going to browse, they're not going to look at ads, they're not going to look at top 10 results, right? they're

21. 9:13

    really about kind of scanning all the options, evaluating them and uh executing the kind of the best outcome that they can achieve. And so this is where kind of a logic and intelligence really meets this marketplace we've been building with intents. Next slide.

22. 9:37

    So agentic commerce really AI is I got to start planning your life and you know be there for context and so really important pieces are private AI because the the most effective AI will be about having all of the context that a person have right streaming all of the conversations all of the interactions you have and really suggesting and being proactive on what can be done and then really not just solving like uh parts of the

23. 10:10

    problem. I think a lot of the AI right now when people try to implement it in enterprises, they're trying to solve like a small use case and I think that's actually the um like a failure mode because what you want is solving the whole outcome. I call it outcome automation where instead of solving like oh you know we can have a better analysis of the emails you can say well actually let's just replace a whole flow and then get this outcome within this parameters we can evaluate easily and now you kind of hide maybe

24. 10:43

    some of the people and processes under the hood into this intentbased architecture. So that's how I see the gentcommerce is really coming this year. It's still going to be, you know, they're going to be manual flows. They're going to be integrating with existing APIs and existing maybe even websites on the back end, right? Where I mean, the example we showed was travel. You're going to type

25. 11:03

    something in. Maybe your browser agent will go and click through Xedia on the back end, but from your perspective, you just got the outcome, the ticket, you want the hotels you want, the museums, etc., right? Even if the museum itself didn't plug in into the system yet. So that's end to end agentic commerce, right? streamlining the whole flow. Next slide.

26. 11:25

    The world is catching on, right? I mean, I've been talking about this maybe year and a half, two years. But the reality is, you know, other people sees this as well. We have X42, which we are partners of. Uh, but also Google just released universal commercial commerce protocol with Shopify, maybe more on e-commerce side on the memory and kind of this like the front end is your uh agent, right? Cloud co-work is really good example and obviously the importance of privacy is more and more

27. 11:55

    right open air and tropic launch specific healthcare systems where they talk about privacy right it's still uh kind of leveraged in the same way right now but there's really an importance and realization that these types of interactions need to have private and kind of dedicated solutions so I still think we have kind of a more complete solution that really covers all the use cases and really prov provides us the fundamental vertical integrated platform through those core pillars we built but it's about now bringing this

28. 12:25

    to market across the board and really bringing this uh to the users. Next slide. So we're moving from this old model where users were product right you know Facebook and Google users are product uh advertisers are the customers to new model next slide where users are kind of at the center right they own their AI and then AI goes and interacts and discovers on their behalf next slide

29. 12:57

    so future is near uh we have real advantage with near intense so let's not miss this opportunity let's keep building keep integrating it everywhere and then let's turn it into aentic commerce by plugging in these components together. And so let's make this year where this vision really comes together and uh goes beyond even web 3 but truly to everyone's device.

30. 13:21

    Let's stay focused and have an amazing year. Thank you. Thank you so much for coming, Ilia. And you heard the man. Let's lock in everyone. Let's move this technology into what the real world, right? Like the borders between blockchain and cryp like crypto and the web tool world are slowly going to vanish and near is going to be at the heart of it. All right, next up we got Bowen Wang from the near one team. Bowen, do you hear me?

31. 13:59

    Yes. All right, Bowen, let me know when to change the slides. Okay. Um, yeah. Uh, great to be here uh in the first uh, uh, town hall of the year. I want to quickly give some updates on the things we're working on. Um, on the near one side, including, you know, protocol like research and chain signatures and so on.

32. 14:24

    Um, so on the protocol side, there are some exciting features we're looking at in the next uh, few months. uh mainly talking about the big ones here. Uh we have um a lot of small uh I wouldn't say small like uh very meaningful improvements from the work we did last year for the 1 million TPS. So we're going to um uh going to go into releases uh in the next one or two release. Um so some of that will be quite exciting including that uh we will be um

33. 14:52

    increasing the function call gas limit um by like three 3x. Um so uh that would be very um that'll be solving a lot of pain points that people have today which not being able to fit um uh enough operations in one uh function call. Uh know that highly requested feature that that is also coming but looking at some of the more uh major initiatives we have dynamic restarting uh coming up. Uh I think um this has been talked about a lot in the past several years but I

34. 15:24

    think um we're finally getting around um to implementing that. Uh so this is um both driven by the maturity of the uh of our development but also by the need that was um because more and more people using near um some like in the past year we have to do quite a few um resharding and there are quite many and there's just like uh a lot of de uh just consumes a lot of development time unnecessarily. So making that uh trigger um working dynamically also um also

35. 15:56

    helps with uh not wasting time there. Um and also we're working on this uh major overhaul of the protocol uh by separating consensus and execution. uh it should uh allow us to have even lower block time um and uh with improved throughput uh and this would also give rise to some uh interesting uh opportunities. Uh Peter can go to the next slide.

36. 16:24

    Yeah. So on the uh on the research side, this would um um that this change of architecture um would make ready for us to um integrate uh zero knowledge proof into the protocol. Uh basically um transitioning the state witness we use today to just a very short uh and and easy to uh quick to verify uh zero knowledge proof. Um, this was the uh starite vision that I laid out in uh near con 2023. Uh, I think uh we're now getting closer to that to that vision.

37. 17:01

    Um, and also this potentially allows um us to have synchronous or like atomic transactions uh implemented on near. Uh those are some um I would say more like uh longer term ideas that uh we'll be working on on the research side. Um but yeah it's very very exciting. Uh can you go to the next slide? Yeah and uh also a quick update on chain signatures uh because yeah with the adoption of intents and the growth uh chain signatures become uh more and more important piece of infrastructure for

38. 17:34

    the near ecosystem. Uh so r uh we have um by the end of last year we've already implemented everything uh related to running the nodes inside trusted execution environment um to have more um uh security and robustness. Uh and now we're going to uh I think in Q1 uh enforce actually enforce the the nodes to be wrong inside the uh trans execution environment. Um and also we have this uh uh new cryptography for ECTS signing that we're uh we have

39. 18:06

    already implemented and um uh it uh improves the throughput as well as reducing uh the latency. Uh so we're going to uh work on rolling that out as well and um um yeah hope to uh scale the number of node operators. Uh this is both um as much of a technical problem as it is a community um governance problem um that we need to uh expand the number of node operators while still figure out how to um you know coordinate

40. 18:38

    between everyone uh making sure the uh the network is secure and so on. Uh and finally uh as we go down this path of uh uh trans execution environment, we know that uh if we rely on um a specific uh type of TE, it may have uh vulnerabilities or have other uh problems. So we're going to look at having supporting hogenous um trust execution environment. Uh so specifically we're going to add AMD support in addition to the Intel TDX support that we have. um so that even if there's a vulnerability or specific

41. 19:14

    problem with one type wouldn't affect uh the safety of the chain signatures network. Uh so yeah overall I would say we have a very exciting year ahead of us and uh let's keep building. Thank you. Thank you so much for coming Bowen. As you heard from Bowen, a lot of great stuff been built already and so much more going into production uh this year. I think it's um super exciting to hear uh what Bone and the Near One team is cooking and also just seeing that while implementing all

42. 19:54

    these insane updates and pushing blockchain technology forward um on a global scale, NIA has never stopped working. So um thanks for that Bone and team. Um keep up the good work. [snorts] All right. Next up, we got Cameron Dennis.

43. 20:16

    All right, Cameron. Cameron. Hey. Hey. How's it going? Can you hear me? Yeah. Beautiful. I'm going to stage. Sounds great. I'm used to being an admin on this thing. So, I was trying to do it myself. But, uh, hey everybody. Uh, my name is Cameron Dennis. I've been building on Near for about five years now. Um and over the last almost two years, year and a halfish, um I've been working on your AAI uh specifically trying to build user-owned AI. Uh this has taken many

44. 20:46

    forms over the year plus or almost year and a half. Uh we built agent frameworks. We've done uh interoper agent interoperability protocols. I've proposed uh DCML which is decentralized confidential machine learning to you know source compute in a much more straightforward way. And my job these days is I'm actually just uh taking near AI to market. Uh working on partnerships, not doing any investments these days. No investments ecosystem. Uh and then uh also somewhat of a janitor focused on compliance and some other

45. 21:17

    stuff. But uh just disclaimer uh everything that is in this presentation we're about to give is not committing us to the exact road map. Things might change. The AI economy is changing every single day and don't really always know where it's going. And so, um, do not hold my, uh, you know, hold me accountable to every single little detail here. Um, but yeah, I just want to give you guys a quick overview. I

46. 21:40

    know I's given the the talk a whole bunch about user owned AI, but I'm going to do it again and uh, talk about inference, some chat, this interesting memory product that we're designing and looking for design partners with, and uh, some shade agents. So, uh, as you've already heard, uh, Bowen is, you know, laid out the the road map for where near protocol is going. This is just the way I like to see near generally is um you

47. 22:04

    know this is why I'm not on the marketing team by the way is I still think near as the open web and that these are the three parts to it and so first is to build the layer one that scales uh can handle can uh decentralize dynamically uh and linearly with demand.

48. 22:21

    So more demand that comes network more decentralized it gets we have to handle that decent that that demand. Thank you Bowen for all the great work on that front. And the second piece is near intense. Uh this is you know checking the blockchain box one last time. Making sure we have uh not only our global state machine to attest you know identity and money and governance to but also work very dynamically with other state machines other blockchains and I think the chain abstraction intense team

49. 22:47

    is crushing it on that front. I think we're about to hit uh 10 billion dollars in volume on Intense which is exciting. Uh but why is this important? as Ilia described, it's all to power the agentic economy and uh really the that equals the entire economy and uh we need to make sure that that AI is very much so owned by the person uh using your near account to use to be that that identity solution. It is that sign on and so uh just take a look back. By the way, this

50. 23:16

    is a presentation I gave early uh like late late last year in December. So, some of these metrics might not be totally correct. Um, and this these numbers are wrong, but we're about to hit almost 10 billion volume, which is exciting. And on near protocol, we have about I think it's like about 40 40 million monthly active transactors.

51. 23:36

    That's pretty cool. Um, I don't think many other blockchains have that. And then now we just need to make sure that well we own AI because as I said and many others uh the whole paradigm is shifting and that you're essentially just going to have this assistant that transacts on your behalf that knows exactly what you like and has the exact payment rails to to engage with the wider economy. And this is all going to be done in a verifiable way, whether it be through uh smart smart contracts,

52. 24:02

    which to me is like the coolest thing about crypto is we can finally program money um and near that sort of coordination layer between it all because we need to move away from the day where we're handing AI all of this data and to a world where actually we hold our data and the AI and all these other applications interact with our AI.

53. 24:23

    This to me is like the the cream of lme what we should all be working towards uh inside and outside of near like if anyone in the ecosystem even outside of near is uh working on this I want to talk to you because I want to sell you private inference and so uh the reason why is because if this whole world is transacting and we're not able to verify the privacy ourselves we're essentially

54. 24:46

    stuck at the whim of a third party another country another company uh telling us what we can and cannot do and you know literally managing our entire digital lives. So we need to verify this privacy and the good way to do this is via the confidential compute. Um as Bowen had mentioned there's this TE trusted execution environment. Uh I like to think of this thing as a blackbox. Um

55. 25:09

    this blackbox you can uh throw a bunch of stuff inside of it. And if anyone's interested in how to do that, uh there is this amazing open source project called DStack built by the FA team and the greater ecosystem where you're able to throw anything that can run in a Docker image inside one of these trusted execution environments, these black boxes. And um what's cool here is you

56. 25:32

    could actually run very large AI models in these TEES that sit across uh like all these GPU cards. And last February, uh, Nvidia came out with the ability to have these TEES on the GPU. Historically, they were always on the CPU, and that's really cool, but you weren't able to run large models um on CPUs. And so now that they're on GPUs, you you can. And what this enables you to do is uh I'm going to skip some of these and come back to it. What you're able to do is, you know, run this model.

57. 26:05

    You chat with your AI and you can encrypt that prompt. You encrypt the prompt, you send it to this TE. Inside this TE, this AI model is running inside of it. The the prompt gets decrypted inside of that black box and then you run the inference. The inference for those who might not know is essentially uh taking your prompt and then asking the AI and saying, "Hey AI, like answer

58. 26:29

    my question." It's going to see all the things that it was trained on to answer that question to provide you that that response. That response is then encrypted. uh inside the TE sent back to the user. The user now has it and uh can decrypt it to see the response. This all is happening by the way in like sub seconds like milliseconds. And uh the really cool piece here is that you get

59. 26:53

    these hardware attestations from the CPU to the GPU back to the user to prove that this thing has been private the entire time and this data is not leaking. And what I love about this is that um to be totally frank, I don't trust these uh large companies saying that, hey, like we're not going to train on your data or it's safe with us. Um even if they do have good intentions, I see a world where uh super intelligence

60. 27:18

    exploits every vulnerability on the internet and leaks your stuff. Um and if it leaks all your stuff, you know, we're going to have a pretty bad time in society. And so, um, I want to make sure that we can have that verifiability at every single step of the process. And this is a really great way to do this. Um, and so you get these hardware attestations. These attestations are like, you know, for the all the crypto folks out there, um, like

61. 27:42

    a transaction hash where you're able to then take that transaction hash, go to a block explorer, search it up, and say, "Hey, this transaction actually happened." And so um now you have a way to prove that your prompt is private which is super exciting and this allows you to do you know it gives you the confidentiality you get the verifiability it's now on the GPU and you can throw pretty much run any

62. 28:06

    arbitrary computation inside. What is very cool and we're uh I don't know how much I should talk about this but I'm going to say it anyway is um we'll eventually be able to run like arbitrary tools inside of these tees. And so think of like any MCP server out there. Um you can actually verify privacy of uh the activity happening using those tools.

63. 28:28

    And so that's exciting. And you know we're not the only ones focused on this. This is going to be a huge uh market and this is from a Gartner report. Uh Gartner's one of the top market research organizations in the world and they're expecting that the confidential compute market is going to be a $50 billion market in 2020 2030. And that is four years from now and uh you just read some of these stats here. It's really impressive. Uh confidential compute is going to be I think the standard for AI

64. 28:57

    around the world and um right now the alternative instead of you know oh the TE based confidential compute uh you have the trustbased systems like that's saying hey my privacy policy says I will not train on your data or you know we it's stored in this safe way. uh but these systems get hacked all the time and you know it we've all seen that sort of fail. The second approach is the on-remise. So you've probably heard a lot about like edge AI or being able to run models on your local device and

65. 29:29

    that's really cool but there are some significant pitfalls here where you're not you're not able to run the best models the biggest ones um and that limits you you can't run many models. So if you run an image model and a text model and a video model that's not really possible if it's all happening just on your device unless you have a really really powerful supercomputer which none of us really have. Um so we

66. 29:51

    have to go call the cloud. We have to go use these external data centers that run all this really you know impressive expensive hardware to run these larger models. And um you know this exists it's cloud confidential computing but you still have to trust the cloud provider and everything's not end to end encrypted. I mean, I actually see a future where eventually those things will be end to end encrypted because there's going to be such demand for privacy and AI. Um, but right now we

67. 30:16

    have this little sweet spot and we're going to, you know, hit it pretty hard because uh there's a huge market because a lot of organizations in healthcare and legal and finance and I don't know if you guys saw I'm just going to actually wing it here. Um, like Sam Walman says that you shouldn't be throwing all your sensitive information into a, you know, uh, a lawyer should, you know, not always use AI for sensitive legal cases because, well, they can get subpoenaed

68. 30:43

    and there's not attorney client privilege between you or your you and your AI model. You have privilege in the legal sense between you and your therapist, you and your lawyer, you and a priest. Um, this does not exist in a lot of models today. So where does that leave me? Um right now this does take about a 5% performance and cost increase and decrease. Um that's marginal in terms of like these models are getting so good and so fast that like people can't even tell. Um and this is why we're focused on building a product. So

69. 31:17

    the first product is the new AI cloud and this is to primarily power verifiably private inference. Um this product is primarily for developers and businesses. I want to emphasize businesses because well they have a lot of users and those users should have verifiable privacy and so uh what we're able to do is run these larger models inside of this trusted hardware. Uh you can you can tell us the models that you'd like us to run but right now it's like the the best open source ones. Uh so GLM 4.7. It's funny if I listen to

70. 31:51

    this later and there's be much better open source models but GLM 4.7 deep seek v3.2 who um we have uh OpenAI OSS 12B uh I think we're you know we're going to consider hosting some some image models and uh then we just give the developer business an open AI compatible API key for those who might not know there's this uh essentially AI an API key that works very very well with any other u model that you've used to using why that's important is because it takes

71. 32:22

    literally two minutes to add it to your your database and so now that you can just add our models that is verifiably private within a minute or two that's pretty cool and you can verify privacy along the way. So this is the first uh product and you may have seen in uh on Twitter you know we're working uh quite closely with Brave and to enable uh verifiable privacy for them. you know, they actually recently hit, I think it was 104 million, uh, monthly active users, you know, bringing this to the

72. 32:51

    masses and absolutely love working with Brendan and Ellie, uh, who are really committed to the future of privacy and or the present of privacy. Um, if you might not tell, I'm I'm using Brave right now. I'm a Brave Power user. And, um, this is a great way to sort of get access initially to some of our models, but you could also go directly to the Near AI chat, uh, chat.near.ai or private.ai. uh AI is a great way to

73. 33:17

    start exploring the models. I use a different browser um for my chats so can't log in but you can just access all the models that we ho that we host and that's for everybody that's the BTOC product um it's pretty it's free right now um we will throttle you if you start using it for a lot of you know eating up a lot of tokens so please don't but you should definitely use it um and that to me is a bit more of a

74. 33:46

    sort of showcase of how private inference works at scale. Um, and how fast it can be like without having to uh, you know, really you can actually prove the attestations yourself in the chat as well. Um, and you know, you get all these benefits of running it inside these models inside of TEES. So yeah, here are the models that we run. Uh, this is how it works. I kind of explained this earlier. Um, so I'm not going to do that again. So this is the the verification widget.

75. 34:16

    So you're actually again able to like take that attestation from the CPU or the GPU and then check it against uh Nvidia or Intel's uh essentially at testistation checkers. What I would love to see making a sort of request for people to build stuff is more third party atstation checkers. I don't know exactly how you're going to go do this.

76. 34:35

    Um I think you're going to need some sort of repository of all the other hardware that was sold by Nvidia or Intel. Um there's a lot of power that sits in Nvidia and Intel here. Um, but I would love more thirdparty uh hardware at the station checkers. That'd be really cool. Not much of a business to build around it right now. I think confidential compute needs to sort of grow more in order to warrant that. But, uh, highly recommend checking that out.

77. 35:00

    Um, and another important piece is if you're running, you know, AI for large enterprises, managing critical data, you need to hit certain service level agreements, SLAs's. And these SLAs's are uh around uptime around support and when we're doing all of this and so a lot of our enterprise customers are quite happy with what we're doing and we are you know assuring that you're going to be

78. 35:25

    able to rely on us for your your kind of most intense use cases. Talked about this a little bit. Um you know I'm not going to dive into to the memory piece right now. If you are interested in uh the being able to move all your context from one chat to another chat or specifically if you're building a business that manages a lot of u context for different use cases that's very sensitive. Um we are exploring a private portable memory product. Uh Google came

79. 35:56

    out with one today. Notion has one but the problem is that this is not portable. You can't move all your contexts from thing to thing. And um an interesting part about this is that you can actually use your near account as the authentication tool to log into uh your private portal memory system. And so we are very interested in exploring this with uh partners who well are are building a lot uh and are willing to pay because we are building a business here.

80. 36:26

    And I think that that is an important flag uh for a lot of crypto people out there that uh we're not issuing another token here. We are actually making money, making profit and doing things with that with that money to benefit the new ecosystem.

81. 36:42

    And so, uh last but not least, this is a framework that's been around for quite a while. It's called the shade agent framework. And this is sort of uh connecting every like everything that's happening in intense land and uh privacy and TEEES where you're now able to run these agents inside of these TEES that have this ability to transact across any blockchain. So uh this work is being

82. 37:05

    done by the Proximity Labs team. Shout out to them. I think they were supposed to talk on this and they're probably going to give a deeper update on the state of shade agents uh maybe next town hall. But um yeah, they've been crushing it on building cool shade agent use cases. Uh they just came out with the and don't want to spoil all the the good stuff, but it is out there the um stable coin transfer protocol, the STP. Um that

83. 37:28

    is I believe leveraging the shade agent framework to move lots of assets from uh one chain to another. Um, and the cool thing about shade agents is that you can arbitrarily plug in different tools, different LLMs, different chains, and this is just a a really clear way to uh well use AI and uh build agents.

84. 37:51

    So, talked a little bit about that. Oh, yeah. And then we're, you know, in terms of the near AI growth strategy, uh going after uh robotics, we think that this is a very interesting space. Um we have written some blog posts about this in the past where um if you're manage if you have a robot in your home and that robot is not able to run the you know the largest most powerful models they're actually taking all your context that's

85. 38:15

    inside of your house and sending it to a cloud provider. That cloud provider is running a model that is then running inference on that data and then giving you back the result to have your robot talk to you or action for you or something. I don't want all of my home data to go to a cloud provider and that inference provider to be able to see everything and the cloud provider to see everything. I would much prefer a cloud that keeps all my data very safe and

86. 38:40

    using verifiably private inference where you get the attestation that you could check yourself is a fantastic use case and so um currently unpacking that you know talking to some healthcare organizations, financial services uh governments etc.

87. 38:55

    But um in order to serve these customers, we need a very high compliance certificates. So we are actually in the process of getting SOCK 2 compliant in ISO 2701. Uh if you don't know what those things are, God bless you. Um if you do know what those are, you know, it takes a long time. It's essentially a a standard and a certificate saying, "Hey, this company is managing your data securely and um

88. 39:20

    you know, that's important to to sell into these organizations." it's actually prerequisite and so um I have a thesis where all these compliance standards although it's very important for access control and making sure that most organizations are behaving with you know good data governance um I see a world where they're not going to matter when super intelligence uh can exploit every vulnerability on the internet and you're

89. 39:43

    going to need verifiable privacy and you're going to need confidential compute as the actual baseline for for privacy and consent and So, uh, that's a little further out. We're not quite there yet. We're not, you know, people aren't running fishing attacks at scale, which, you know, are destroying society and, you know, we're not seeing all the data breaches. Although, there was a big Instagram one recently. Um,

90. 40:07

    you know, I want verifiability. I'm not cool with the, hey, trust our privacy policy and, you know, with the power of super intelligence. And so, super grateful to be working with organizations like Brave. uh they've been a fantastic partner and really look forward to, you know, driving AI adoption with them and other tech companies as well. Um you know, Brave is not the only one who cares about AI and privacy. There's many many of them out there and uh we do plan to work with

91. 40:32

    everyone who wants verifiability. And so that's pretty much it. Um this is normally what I use in sales decks, but uh this is how you can secure your AI workloads with me and your AI. um happy to dive into uh tech if you're a business or enterprise that you know generates lots of tokens. Um happy to talk with you. What I'm not doing is I'm not connecting with everyday developers who just want to uh test it out. You're able and super encouraged to check it

92. 41:02

    out, but just go to cloud.neer.ai or near.aicloud and just get an API key. You don't need to talk to me. Um you can talk to the telegram chat. There is one for support, but um it's all super straightforward. If you have ever used any AI model before in any API key, it's the same exact thing. So, we've made it so simple, we don't need to provide a ton of support unless you're an enterprise looking to pay a good amount of money for tokens. So, that is the gist of what

93. 41:32

    we're building today. I'm not going to dive into what is going to come out, but that will come at the right time. Uh, that's all I got and thank you for having me and you can follow me at Cameron_Dennis on Twitter. Um, message me on LinkedIn if you have any questions.

94. 41:52

    Thank you. Thank you so much for coming Cameron as always a pleasure. um such a thorough walk through of everything that is cooking within the near AI um side of things and super happy you came today cuz I think a lot of people have been asking that question like what is actually going on but ladies and gentlemen there you have it a ton of stuff is being built and there's an active effort to on board enterprises

95. 42:25

    and as it seems from the got report the private AIN industry is only going to grow. So seems near protocol was is just right on time. All right. And last presenter of for today's town hall is Stephanie from the events team and she's going to walk us through all of the great events that are coming throughout the year. Stephanie, welcome to the show.

96. 42:54

    Awesome. Thanks so much, Peter. I appreciate being here. Uh, next slide, please. Awesome. So, we're going to kick off the year with a near intense day global series. Um, starting with an online version of it on January 22nd. So, tune in. I'll give you some more details on the type of content that will be taking place that day uh in a little bit. But just so you guys know, for ETH Denver,

97. 43:17

    we're planning for February 18th. And then for etc token Dubai consensus Miami and token Singapore we're still uh working on announcing the dates but uh please uh take note that we will have a presence there. Next slide. Um as I mentioned I'm going to give a dive a little bit deeper into the in near intense day online. So we have four awesome panels that were helped to be curated by uh Bobin and Harshett. So big shout outs to both of them. uh we're covering privacy uh apps and intents defy and then VC and market structure

98. 43:49

    and hopefully we get to secure a special session with uh Kane from Infinex. Uh the goal of this is to not uh to basically keep our foot on the gas and make sure that people keep talking about Near Intense know um what it's about and also to highlight and showcase our partners who've been helping um spread the word about Near Intense. Uh next slide please. Nearcon. Nearcon is just

99. 44:12

    around the corner. I know it seems far away, but it's really really not. Uh it's taking place February 23rd and 24th in San Francisco. U some just highlevel numbers. Uh we're pacing really really well so far. Almost 700 registered, 530 accepted, 400, excuse me, 157 still pending. Um and just to get some context of the type of audience that we're able to pull in, mainly senior builder and

100. 44:35

     about 50% of it is founder/ seuite with a strong operator background. So really, really good crowd, really excited. Um, which just kind of signals we made the right choice and uh moving content San Francisco from a production and logistics standpoint. Um, production is fully underway. Timelines are locked and we have we don't see any material risks to the project. Um, and then just the highle overview of what we're planning

101. 45:00

     for uh for that week. Sunday we're planning an uh near foundation and ecosystem dinner. Monday is near con day one followed by a speaker reception and then Tuesday is near con day two followed by closing party. Um high level in the agenda and programming um I always think that the agenda needs to accomplish two things. Explore the big universal questions shaping the AI economy as well as position near as the infrastructure for that AI economy. Uh

102. 45:26

     we have two stages this year uh from 11 to 5:00 pm and then from a sponsorship and media perspective we have uh six sponsors perceive0ero venai sovereign AI rhea and brave so shout outs to those partners for coming in we also have a media partnership so they're going to be on site doing live um presentations with the speakers right off the stage really exciting stuff there and then um like similar to last year Global is our PR agency and they've already hit the ground running securing the Financial

103. 45:56

     Times as well as CoinDesk to do some coverage for the event. So, really good stuff. And then finally, we this year we opened up a community partners program. And so, we're collaborating with local organizations in the Bay Area, including um the uh Silicon Valley AI hub, um Mission Bit, House of Web 3, Cooperative Futures, and Frontier Towers. And we're still taking more uh community partners.

104. 46:18

     So, if anyone in the ecosystem knows folks that are interested um local organizations interested, uh please send them my way. more than happy to take that call. Next slide, please. And then an events roadmap. I mentioned we're doing near intense day, but also wanted to highlight some other events. We're going to be we're going to have some kind of presence at um Nvidia G GTCAI digital asset summit and as I mentioned before ECC um we're going to be

105. 46:40

     sponsoring middle Asia. We're going to be at token 249 uh Dubai as well as consensus Miami. And I next slide. Uh that's it for me. Thanks Peter. Appreciate your time. Thank you so much for coming, Stephanie. So, you heard what's up. If you're in the San Francisco or the Bay Area or just in any country nearby, make sure to sign up for NEON and head over to the great event that's going to be thrown here at the end of February. If you're

106. 47:12

     not able to participate, Nia will be present during most of the global and biggest events uh around the world during the year. That's a solid lineup. Stephanie um couldn't look any better. So, kudos to the events team for um throwing all these things. Lastly, guys, I just want to take you through a quick recap of what happened in 2025.

107. 47:38

     And [snorts] I'm going to kick it off with 28 reasons you should be bullish on near. This is a post published or a video published on Twitter with the infamous Aussie Mendias uh walking you through all the reasons why you should be bullish on near. If you haven't watched it, I highly recommend it. I'm going to give you the TLDDR so you don't have to watch it all even though I do recommend it. The TLDDR is that Near isn't just positioned for one narrative with proven product traction.

108. 48:11

     Near is positioned at the intersection of three areas. AI infrastructure, crosschain liquidity, and privacy. So, win one, benefit from all three. Let's go. Next slide. And if you haven't already, I highly recommend joining the Neil Legion.

109. 48:34

     This is where the beat of Near is at the moment. As a member, you can join the Eco Collab Core call and get a peak behind all the scenes that's going on in Near. You can arrange and participate in IRL local events in your city or country to get the Navana vibes going and connect with all the players of the ecosystem. There have been two recent events uh and more are coming. So check out the near legion the map on the near

110. 48:59

     legion page to see if any local or any people in your city or country are around and connect with them. Get the vibes going. Next one up. So just going to take you through a brief uh 2025 and 2026 outlook. This was a report article posted by the Neo Foundation at the end of the year and I'm just going to walk you through um three highlights um that I think were essential to NEO protocol last year. So

111. 49:32

     obviously the near one team achieved 1 million transactions per second using affordable consumer grade hardware and you can verify all of this stuff in the article that the team published. This is a huge milestone for near but also for the blockchain space in general and it demonstrates that a shed infrastructure that can that can scale horizontally for high load apps and well beyond the throughput of traditional payment networks is going to work and it's being built on near.

112. 50:09

     Another thing to note is also that finality now occurs in 1.2 few seconds making near one of the fastest settlement layers in the industry and as you saw Bowen present earlier on they're just going to keep pushing the boundaries of what you can do with this with the near technology so kudos to to to Bone one and the new one team. Second thing I want to highlight is of course the near intense stats. So in 2025, Near Intense became

113. 50:43

     the fastest growing crosschain infrastructure, its um unifying liquidity and it's universal execution across markets from D5 to AI. And I'm not sure if you you probably have noticed that intense has be is becoming a widely adopted infrastructure layer for seamless crosschain transactions. Um counting more than 28 uh chains and I think it's more than 25 major projects that have integrated the technology.

114. 51:13

     um the likes of Ledger, Kyber, uh Infinex, list goes on, are already using the technology. And looking at the stats, it's pretty amazing. You see, we're almost at 10 billion um total volume with two million two billion of those occurring in the last 30 days. During 2020 or since inception, we've had more than 15 million swaps and more than 500,000 pe unique users use this technology on a monthly basis. And the beautiful thing about it is they

115. 51:51

     probably don't even know they're using it. It's just doing its thing in the background, raking in all of those great fees that the protocol has started to produce that I also mentioned earlier will only increase. So near is also now a revenue new generating machine. Big kudos to the intense team. I cannot wait to see what 2026 is going to give us. [snorts] Lastly, of course, we had the launch of the near AI cloud and the private chat that Cameron just walked us through. And um these two products have introduced a new category

116. 52:30

     of privacy first intelligence for both enterprises and consumers. So developers can now deploy models that keep user data encrypted end to end. And it this solves as also was laid out and the thesis that lit up with Ilia earlier one of the most significant challenges in enterprise a adoption but also in truly private AI adoption i.e. your assistant that runs on your phone should

117. 52:54

     be your brand. It should not work for be someone else's tool. It should be your tool. you should be able to own that and you should be able to verify that and that AI should be able to transact privately on your behalf. Um, one thing to note about this is of course that the near AI infrastructure is now supporting deployments serving more than 100 million users as Cameron just pointed out through integrations with Brave Open mind which is very very

118. 53:24

     interesting robotics company out of San Francisco. I highly recommend uh you to check it out. And of course we have the next generation travel management platform Trevi which also leverages near intents. And I think as Ilia pointed out, 2026 is going to be the year where we start to see the convergence between near intense and um AI technology and it's going to be truly truly exciting to see what kind of products that are going to be built

119. 53:54

     putting the technology into real world use cases, not just keeping it to swapping assets, which of course is great. Um, near intense has also uh a plans of making any digital asset or any asset tradable. So stocks probably coming up as well.

120. 54:16

     That's it friends. Thank you so much for watching today. The future is near and as you've seen today, Near is shipping infrastructure for for AI, but also for the next wave of commerce for agents and for the next era of the internet. And um this team won't stop until near is the backbone of it all.

121. 54:40

     Near is where users own their AI. It's where a agents coordinate and the aentic internet scales. As you've seen, this community is grinding, building, and showing the world what's possible with blockchain technology. I want to give a huge shout out to every developer, founder, and community uh contributor pushing this vision forward.

122. 54:58

     NAR isn't just a protocol. It's a movement for truly private userowned AI. As always, what is was presented on this uh stream is not financial advice. Do your own research. Use your own judgment. But if you're paying attention, you know where the signal is. Subscribe to Near to stay in the loop. And if you're building something amazing, this stage is for you. Do not hesitate to hit us up if you have something you'd like to present at the next town hall. See you next month.

123. 55:29

     Keep building, keep grinding, and stay near, friends. Have a good one. Thanks for watching. Truly open AI is near. Follow near protocol. Subscribe to Near Week.