Eduardo Soubihe

@latitudesh @consensus2026 let’s go!

3rd year at @consensus2026 … guess that makes us OGs - booth #3914 - New locations (spoilers: 🇦🇺 & 🇨🇦) - k8ns in Public Preview - Solana Seeker giveaway - builders side event > pull up 👉 luma.com/thjwjiie 𝙄𝙛 𝙮𝙤𝙪 like 𝙩𝙝𝙞𝙨 𝙩𝙬𝙚𝙚𝙩… 𝙮𝙤𝙪’𝙧𝙚 𝙖𝙥𝙥𝙧𝙤𝙫𝙚𝙙 𝙛𝙤𝙧 𝙤𝙪𝙧 𝙨𝙞𝙙𝙚 𝙚𝙫𝙚𝙣𝙩 👀

@__Achille__ @usefiretiger all your points are moot if agents continue improving even at modest rate

I heard in several conversations that coding agents are the end of open-source software. I strongly disagree.

Running a company: 2020: can you survive a pandemic? 2021: still here? we’re going to give all of your competitors $100m series A rounds. 2022: wow, you made it? okay, all engineers cost $600,000/year now. 2023: nice job! okay, SVB failed and we’re going to take away your bank account. 2024: a survivor I see. but can you pivot from ai to crypto to defense tech back to ai-enabled defense tech in a 12 month period to stay relevant? 2025: unfortunately all of your competitors have raised $2b series B rounds. oh and only 500 engineers are relevant and they cost $100m/yr each. 2026: well, well, well. you’re still in business? let’s deploy the thunderclap of godlike LLMs from the heavens so all of your customers can rebuild your app in 2 hours. can you survive?

Most Ethereum teams upgraded to cloud VMs and called it a day. But VMs come with hidden costs nobody talks about. Victor Chiea from @latitudesh is on the Redford Stage right now breaking down what Ethereum workloads actually need 🔧 📍 Redford Stage | 12:50 - 13:10

@dexhorthy I want my dev and agent environments to be exactly the same, here’s a project from latitude that plans to do just that: devbox-sdk.dev

lot of sandbox infra is building for the "cattle" use case but I think we'll see just as much use cases for "pets" as background agents become the norm. I still find it hard to shift from "agents running in remote compute" to "agents running on my laptop" - not because of the agent or the inference, but because I'm used to having everything right in its place, libraries, clis, repo paths, etc etc. The first step to remote coding agents should feel at least somewhat continuous in this regard You bring the server or base image and configure it how you want with API keys, repo checkouts, etc - then the agent platform runs sessions on that host sure you need some more plumbing to make it work in terms of compute efficiency but the cost of a small-to-medium EC2 is pennies compared to the cost of tokens I'm blasting through this thing. enterprise side here interesting too...run everything in your VPC etc.

@mehulmpt when that doesn’t work, reach out to us @latitudesh

genuinely insane release. all those hot AI sandbox startups you see, every single one of them use firecracker (or similar stuff). > get $$$$ aws credits > build this AI sandbox startup > use non-bare metal instances with custom upscale/downscale infra > ???? > profit

✅ We are now SOC 2 Type II certified! Everything you love about our platform, now with enterprise-grade compliance.

.@conductor_build replaced Cursor and Github Desktop for me while making me considerably more productive. Cool product

@HeyGarrison yes but DeSiGn

let me guess - it's a chat box?

@iximiuz @latitudesh when you outgrow them we'll be here 😁

@esoubihe @latitudesh Sorry, you lost me. I keep looking at how AX102-like machines are x4 on your pricing page, and can't really think of a single reason to switch.

Just paid over 600 EUR in setup fees to Hetzner. AMA x3 monthly price to just set up a server is nuts

@iximiuz @latitudesh Here's an image from our internal pricing tracker. Roughly same price for large machines. The smaller servers that cost 80% more have 20x more throughput and ~2.5x more performance than theirs. Also LSH's 23 global locations, global network, programmability etc

@esoubihe @latitudesh Hetzner bare metal provides the exact level of control I need. Anything more managed would be unwanted and anything less managed would be inconvenient.

@iximiuz @latitudesh and you can get even cheaper by buying it directly from dell and running it yourself. doesn’t make it better

@esoubihe @latitudesh Have you even checked Hetzner prices? Your offer is x5 more expensive. Even with the new setup fees, Hetzner is a much cheaper way to get bare-metal servers.

@ovaistariq interesting ripple effect of sandboxing

@esoubihe Agents need isolated workspaces per-run for things like storing artifacts, persisting checkpoints, tool outputs, etc.

Yup. There are assumptions baked into platforms that are breaking. One example I am seeing is that traditional storage platforms expect relatively smaller number of buckets compared to the number of files. But with agentic workloads we are seeing an entirely different pattern - millions of buckets each with very few files.

Kicking off @NVIDIAGTC 2026! Come say hi to our team at booth #3222 and grab some cool LSH swag 😎

@bosunski @jmsalcido @levelsio The reason this is a problem is because tool calls are made from within the model. What you need is an isolated sandbox with secrets. Agent calls sandbox -> sandbox returns structured data to agent

@jmsalcido @levelsio You tell me how an agent can take action against other systems, on your customers behalf with the said alternatives without over engineering.

I wonder how the said AI will call the APIs behind authentication of all sorts? Will you also give them API tokens?

Here’s how this plays out. Software used to be too expensive and hard to write to automate most things. Now it’s vastly cheaper and faster to code. Thus, leverage has gone up dramatically, which means we’ll use software for far more. Leasing to more demand for engineering.

Adopting AI into your work makes everyone a CEO, and there's something about it that very few people understand (mostly because few people are CEOs): When you have an front-line job, you spend maybe 1-5% of your brainpower of high-level, critical-thinking strategy. Most of your job is just running some SOP that's been given to you. Move up a couple levels, maybe you're managing a little team, and you might spend 10-25% of your time making critical strategic decisions. The rest of the time, you're executing departmental strategy that was handed to you, and your time is spent making sure your team carries that out. Once you're the CEO, you'll have (if you're good at your job) delegated all of the routine problems and issues that are straightforward to solve to capable executives. What's left for you? Only the hardest and most critical decisions. The better and more capable your staff, the harder the questions will be that bubble up to you, because they take care of everything else. Those become the only duties you have left: thinking REALLY hard about very dfificult, ambiguous, strategic decisions. And now it's your entire job. Instead of 10-25% intensity (or less), it's 80-90% intensity. Incidentally, this is why you hear about CEOs having these intensely regimented lives and health-oriented habits: it's all designed to biologically support the fact that their brains have to be operating at peak capacity nearly all the time. So now everyone is starting to manage armies of agents doing the routine parts of their job. If you're good, you can distill your job into a clear SOP that the agents run, and now "all you have to do" is oversee them... ... and now lots of people are learning that being the boss isn't quite as easy as they thought.

@gregpr07 would love to support your compute cc @latitudesh

The infra god has spoken