Gio Logist (@ItsGioLogist) header
Gio Logist (@ItsGioLogist) profile picture

Developing, Designing, Dreaming. Thinking. 20+ years in tech. Coinbase Alumni. Ai & GPU Infra @b3labs

San Francisco
Joined February 2011
Photo by Gio Logist (@ItsGioLogist)Photo by Gio Logist (@ItsGioLogist)Photo by Gio Logist (@ItsGioLogist)Photo by Gio Logist (@ItsGioLogist)Photo by Gio Logist (@ItsGioLogist)Photo by Gio Logist (@ItsGioLogist)Photo by Gio Logist (@ItsGioLogist)Photo by Gio Logist (@ItsGioLogist)Photo by Gio Logist (@ItsGioLogist)
This was a fun experience to build. I think sometimes we take all that goes into hardware so for granted. With B3IQ now and our warehouse in Eugene, we’re constantly reminded of all the nuances that go into these nodes and tinkering to make them better 🤙🏻🙏🏻
Everyone’s talking about NVIDIA, but almost nobody can explain what its GPUs actually do. Check out our interactive guide to how AI compute works, from data centers all the way down to the atoms inside the chips. ➡️ b3iq.org/studies/to-scale
3
8
387
Quick lil iOS widget app for personal use, because my dumbass kept forgetting about trips I planned months ago Surprised Google Calendar can't already do this. But these are the types of things that are great to one-shot w an agent. Years ago it woulda been one of the things I have to put on the backburner while working on larger projects 🤪
2
85
He did tell us the plan
Replying to @wintonARK
Revenue growth is insane. Never seen anything like it.
77
278
24
9,135
627,117
Gio Logist retweeted
Be your own Neocloud at b3iq.org.
1
9
3
59
7,018
Agreed. It’s an interesting experience, for sure. No longer capped at your words per minute or ability to test and ship quickly I’m def excited to see what happens from “non tech” people as well though. Eng was previously the blocker, in a lot of companies. A lot of that blockage has disappeared, and so good and creative ideas can get tested and spawn at random. I’m here for it ✊🏻
It's such a weird experience that I can now build things faster than I get new ideas for things
100
Gio Logist retweeted
how it feels on Sunday when that usage limit reset hits
3
1
14
632
Gio Logist retweeted
People have no idea the transformation that we've gone through as a company, truly mind blowing. Started in crypto gaming, acquired an industry leading gaming systems integrator, scaled that business + launched our AI / compute business recently. Many failures along the way, but never bet against ex-@coinbase founders. We never gave up, kept building throughout the bear on something real. Now @b3labs is solving complex problems + scaling to meet the massive success & demand we're seeing in AI. Our ambition is to democratize compute: distributing ownership of H200, B300, 6000s & future blackwell + latest gen AI hardware to more individuals... This speaks to the ethos of crypto (democratize finance = democratize power = democratize intelligence), and most are sleeping on what this entails for the base:0xb3b32f9f8827d4634fe7d973fa1034ec9fddb3b3 token & our community. We hope you'll join us on this mission.
21
13
2
88
5,719
Gio Logist retweeted
Buying a full H200 node outright is out of reach for most individuals, so B3IQ lets owners start with an initial payment and pay the machine off over time. We host it in our secure Oregon warehouse. When you don't need the machine, you can opt in to rent it out to labs and research teams, and a share of that revenue can go toward the purchase price. @viktoriya0x walks @CamiRusso through one scenario, a node rented at full capacity for a year, and what compute rents for on the market today.
9
8
4
83
9,555
Gio Logist retweeted
In truth it was spiritually tiring to work in consumer crypto at times because a lot of it is extractive and basically forms of casinos. But since we pivoted to AI & compute with @b3labs - the whole team is energized with the mission of democratizing compute. To me and many others this was the core mission of crypto originally in form of democratizing finance, and I feel the same way about the impact on the world when it comes to compute. We live in a new age where you can convert energy into intelligence at increasing higher and higher ROI, and that’s an exciting world to help usher in and try to distribute that power evenly.
I'm very bullish on anyone offering dedicated on-prem VPC
6
7
2
35
4,806
A couple of weeks ago we doubled down on something that used to be natural and expected on an eng sync, PR reviews. I remember at Coinbase, we would constantly ask for people to break down PR's into smaller chunks, for the sake of quality control. And it made sense at the time. Rather a focussed review vs a rubber ducky stamp. IMO more important now than ever. I think it's still important to trust your LLM's but also audit. If your code-base if filled with terms like "load-bearing", "envelop", "counsel", etc. IE: words people dont actually use, how the f are you gonna navigate? And a lot of people are solving PR reviews w another LLM. Deepseek, coderabbit, etc. Which is great for catching technical glitches But again, how the f are you gonna navigate a code-base w words you don't understand? This brings me back to a thought i've always had while working w folks in tech. If you can't break down your thought patterns and goals into a simple term, it's likely that you don't understand it fully or the impact it has in the real world. For hygiene, you should constantly audit your code-base for slop, make sure your eng team knows what's actually going on (and more importantly, why) vs just accepting a green-checked PR w passing ci/cd actions. This will help everyone involved if/when shit hits the fan and someone has to answer "why? and where do we fix?"
1
143
hey Astra, how fast are you really at using a computer?
Readers added context they thought people might want to know
The video shows Astra executing a pre-planned batched sequence of mouse clicks, keyboard presses and timed waits, not live real-time high-speed control. The poster clarified hybrid inputs and advance planning in replies. x.com/victornunez/st… x.com/victornunez/st… x.com/victornunez/st… openai.com/index/gpt-6-as… openai.com/index/computer…
314
672
291
9,910
1,952,428
I was half asleep last night while working on a new personal project w @OpenAI 's new Astra, which my Chief of Staff on @bot had context on. So instead of over engineering another LLM <> LLM communication pipeline, I just made this simple file and told them to monitor over a loop and reach a consensus. Worked like a charm and woke up a bit later to a decent plan w their shared context haha. Just a quick lazy hack
4
156
For the love of all things, can y'all add a rule to your Claude.md to persist tabs/filtering/item-selected/etc state in URL. The amount of times I've gone to open a new tab on an app and the state hasn't carried over. Or bookmarked, same. This is not a hard problem to solve 🤦🏻‍♂️
2
130
Gio Logist retweeted
Little piece of advice: Buy enough compute to privately run >= opus 4.8 level intelligence offline, in your home before it’s too late.
7
7
1
56
7,473
Gio Logist retweeted
Where does your proprietary data go once AI agents start touching it? @yorkerhodes, Director Digital Transformation, Traceability & Co-Founder of Blockchain at @microsoft, describes a realization spreading through the enterprise space, and what it means for anyone with a proprietary way of doing business.
Introducing B3IQ. A new model to own, run and monetize compute. Already being used by faculty, researchers, and student teams at Stanford, NYU, Penn, UChicago, Dartmouth, Waterloo, and the University of Hawaiʻi.
89
12
7
132
42,929
Gio Logist retweeted
The agent you use for crypto shouldn't make you go somewhere new to use it. B3OS has a generative UI, an MCP server, and a natural agent interface, so it shows up in the tools you already have open. The plan is to plug it into iMessage, Slack, or any channel where your other agents already live, where it handles the onchain work and coordinates with the rest. @seangeng on how B3OS meets you where you work
6
2
31
4,789
Gio Logist retweeted
Where does your data go every time you prompt a hosted model? B3 CEO @darylX24 on why that answer is starting to matter, and why local inference is becoming the default.
4
6
29
2,048
100%. There's a clear difference between using an LLM w specialized tooling on top vs just an LLM directly. And Caddie does a great job of showing that 🤙🏻
In trading, stale data is just a slower way to be wrong. - Caddie reads the market live the second you ask. ETF flows, liquidations, whale moves, fear & greed, all current, synthesized from every reliable source. - But reading is just the start. same chat, you act on it: set a price alert, build a workflow, swap, hedge, automate the whole thing. The live data wires straight into onchain actions, wallets and gas already first-class on b3os.org. - General models are great at a lot of things. Staying live on the market is a different matter, and Caddie is built for it. right tool, right job.
3
252
Figuring out what a user actually wants when they're talking to your LLM agent is more of a PITA than you can imagine. We spent months on this with Caddie, our AI workflow builder at @b3dotfun. Started with regex. Ended up with a 6-phase hybrid classifier: 1. Pending plan? (deterministic) 2. Last LLM suggested a plan or asked a question? (deterministic) 3. User asked a question with no workflow intent? (deterministic) 4. Editing an existing workflow? (deterministic) 5. Domain keywords or action verbs? (deterministic) 6. None matched? LLM fallback. Our deterministic guards capture ~95% of intent. No latency, no tokens burned. The LLM only fires when we genuinely can't tell. Here's a visual. PS: Curious how others have handled intent classification. I'd imagine it's usually a hybrid. What LLMs downstream are you finding most performant for this specific use case?
8
8
14
1,747
Definitely one of the hardest challenges working w @b3dotfun on B3OS - Due to LLM's being non-deterministic by nature, we had to ride a thin line between allowing our models to shine while at the same time not giving them enough rope to hang themselves with - Training it to use specific domain knowledge we acquired over the years (especially while working @ Coinbase) helped a ton here - In later iterations, we immediately introduced concepts like "recipes", pre/post processing & more. Which made Caddie much more deterministic and reliable, quickly making it something we felt confident trusting our $ in. Most if not all of B3OS employees have used Caddie directly to execute trades, for example.
AI agents are scaling fast; @METR_Evals reports their task completion horizon doubles every 7 months. But there's a massive gap between an agent completing general software engineering tasks and an agent managing your onchain treasury. In crypto, where you're dealing with programmable money, "long running tasks" are often just a slower way to get exploited. With B3OS's Caddie 1.1, we introduces a new standard: Verifiable Onchain Autonomy 🏋️Human-in-the-loop: You approve the plan in plain text. ✅ Deterministic Execution Engine: Every action is a trusted and battle tested explicit node 🔐Absolute Control: Every input + result is logged, and wait nodes pause until you sign. Hear our CTO Sean explain the architecture:
5
164