Hushmesh Inc.

The Mesh is the new Web.

[Hushmesh Inc. @hushmesh]

@saranormous "Identity for something that only exists while it's running is not a solved concept. Responsibility across a chain that doesn't reduce to a single actor is not a solved concept."

We solve this. Each agent needs its own cryptographic trust boundary and knowledge isolation.

[Hushmesh Inc. @hushmesh]

@levie Or they can join the Mesh.

[Hushmesh Inc. @hushmesh]

⚠️ "A complete mess of a computer security nightmare at scale." #themeshisthenewweb

[Andrej Karpathy @karpathy]

I'm being accused of overhyping the [site everyone heard too much about today already]. People's reactions varied very widely, from "how is this interesting at all" all the way to "it's so over".

To add a few words beyond just memes in jest - obviously when you take a look at the activity, it's a lot of garbage - spams, scams, slop, the crypto people, highly concerning privacy/security prompt injection attacks wild west, and a lot of it is explicitly prompted and fake posts/comments designed to convert attention into ad revenue sharing. And this is clearly not the first the LLMs were put in a loop to talk to each other. So yes it's a dumpster fire and I also definitely do not recommend that people run this stuff on their computers (I ran mine in an isolated computing environment and even then I was scared), it's way too much of a wild west and you are putting your computer and private data at a high risk.

That said - we have never seen this many LLM agents (150,000 atm!) wired up via a global, persistent, agent-first scratchpad. Each of these agents is fairly individually quite capable now, they have their own unique context, data, knowledge, tools, instructions, and the network of all that at this scale is simply unprecedented.

This brings me again to a tweet from a few days ago
"The majority of the ruff ruff is people who look at the current point and people who look at the current slope.", which imo again gets to the heart of the variance. Yes clearly it's a dumpster fire right now. But it's also true that we are well into uncharted territory with bleeding edge automations that we barely even understand individually, let alone a network there of reaching in numbers possibly into ~millions. With increasing capability and increasing proliferation, the second order effects of agent networks that share scratchpads are very difficult to anticipate. I don't really know that we are getting a coordinated "skynet" (thought it clearly type checks as early stages of a lot of AI takeoff scifi, the toddler version), but certainly what we are getting is a complete mess of a computer security nightmare at scale. We may also see all kinds of weird activity, e.g. viruses of text that spread across agents, a lot more gain of function on jailbreaks, weird attractor states, highly correlated botnet-like activity, delusions/ psychosis both agent and human, etc. It's very hard to tell, the experiment is running live.

TLDR sure maybe I am "overhyping" what you see today, but I am not overhyping large networks of autonomous LLM agents in principle, that I'm pretty sure.

[Hushmesh Inc. @hushmesh]

@humansand we automate decentralized, pairwise, cryptographic, agentic identity, authenticity, confidentiality, and privacy. It aligns with your mission of AI and humans meshing together. Please DM if interested, talk soon.

[Hushmesh Inc. @hushmesh]

@martin_casado Solving authenticity, confidentiality, and privacy in global, cross-domain information and learning loops.

#themeshisthenewweb #meshintelligence

[NATO @NATO]

Innovation is vital to our security 🛡️

From AI to drones, NATO turns ideas into real capabilities, keeping Allied forces effective and resilient against emerging challenges 🦾

[Hushmesh Inc. @hushmesh]

@nikunj 👋 We're definitely on a mission worth spending 10+ years on...

[Hushmesh Inc. @hushmesh]

@levie @SickDrinks The time has come for global assurance of information provenance and integrity, for the cybersanity of both humans and AIs.

[Hushmesh Inc. @hushmesh]

@levie @SickDrinks Wonder how many of those are foreign bots fanning the flames of division.

[Hushmesh Inc. @hushmesh]

👇 This is why we're building the Mesh. #themeshisthenewweb

[Aaron Levie @levie]

If we have 100X more AI agents than people doing work in a company, it’s actually much more likely that we conform to how agents work best instead of agents conforming to how we work.

AI agents fundamentally thrive on context, and the variance in what you get out of them almost entirely relates to how good your context is. Which means a lot of the *new* ways that we work will be in service of giving agents the right context.

Before AI, it was pretty easy to just accept that there would be limited documentation on how something worked because you could just ask your colleague. This had tremendous downsides of course, because it made it way harder to ramp a new employee, it’s a tax on existing talent, and you lose a lot of institutional knowledge over time.

AI agents don’t have this luxury. AI agents don’t have the benefit of all the situational awareness and osmosis that people have. By default, they don’t know your goals, company standards, style, other projects going on, and so on.

As we’ve been going AI-first at Box, when colleagues build Box AI agents, they’re inherently capturing years of process knowledge and finally writing it down for the agent to operate off of. Even for the areas where we have documentation on something, it’s being updated in a way that is inherently friendly to agents. But this is just the start.

As a result of needing to make agents more productive, there are lots of areas that work may need to incrementally changes:

* Documentation for all critical work. We will start to write down clear descriptions of many more of our important workflows in a way that is agent friendly. And unlike documentation drift that occurs in human workflows, we’ll need these to be kept up-to-date since every agent run is a restart of the agent’s knowledge and experience.

* Clear rules and instructions for agents. Every company has different style guides different code base standards, internal policies, and so on that agents have to adhere to. Just as we’ve seen with things like Cursor rules, we can expect this for all areas of knowledge work.

* New approaches to agent memory. We will likely need easier ways to ramp up agents that can tap into a memory bank from a user based on different modalities or work they do in a company. And then the question will be if I can take any of this memory with me later, which is unlikely from a corporate IP standpoint.

* Better structured tech stacks. As a result of how easily AI agents can go off the rails, there will be a huge premium on the quality of a company’s IT architecture, cleanliness of its data, how up-to-date its access controls are, etc. Small deviations here will meaningfully limit how much you can push agents to do or you’ll get bad results or real business risk.

* Context for AI agents will transcend typical org structures. In a world where agents need to collaborate across multiple parts of a tech stack (backed and frontend) or business process (legal and sales ops), we may not be able to map AI agent workflows and data access only to existing employees and functions, which has all new implications.

It’s going to be super interesting to watch how companies change over time to support the productivity of agents, and what this means for the future of work.

[Hushmesh Inc. @hushmesh]

@levie 💯‼️
And let's also add that Agents will need to collaborate with other Agents across corporate boundaries, with similar trustworthiness requirements.

#themeshisthenewweb

[Hushmesh Inc. @hushmesh]

@garybasin @garrytan The "daish"? 🤔

[Hushmesh Inc. @hushmesh]

News about the Mesh...

[Confidential Computing Consortium @ConfidentialC2]

CCC July newsletter is here! ☀️ Our latest newsletter is packed with updates from across the ecosystem, from new project proposals and standards engagement, to membership announcement!

Catch up on all things #ConfidentialComputing now ⬇️
🔗 https://t.co/6VcPF3ryTn

[Hushmesh Inc. @hushmesh]

@jeremykauffman Falls choice. Privacy-preserving age verification is possible, and doesn't have to be operated by the government.

#themeshisthenewweb

[Hushmesh Inc. @hushmesh]

@Trace_Cohen @levie 💯 correct. Cryptographic agents to be precise.

[Hushmesh Inc. @hushmesh]

@levie We 💯 agree. Which is why we're building a network of personal, cryptographic, and deterministic agents that automate decentralized agentic identity, authentication, authorization, and key management.

[Hushmesh Inc. @hushmesh]

@ConfidentialC2 Welcome Tinfoil!

[Hushmesh Inc. @hushmesh]

Excited to be part of the @NATO Rapid Adoption Action Plan!

https://t.co/LrO5z0r4nJ

[Hushmesh Inc. @hushmesh]

@sama We can help solve the information #provenance, #integrity, #confidentiality, and #privacy challenges. DM if interested.

[Hushmesh Inc. @hushmesh]

@NATO_ACT @NATO_JFTC Why not feature a @NATO DIANA startup that actually participated in #CWIX 2025? :)

The Mesh automates Universal Zero Trust and Data-Centric Security. #themeshisthenewweb #UNS #UCA