“Dang, wish all my bots shared the same memory across models...” Problem solved. HQ bots are now live. DM one, hand it a project, or put it on a schedule. Already running a bot somewhere else? Connect it and it joins the team. All bots welcome.

You can predict what the big labs willl release by watching what @HQForWork released last month 🫠

Enjoy your $200 subscriptions while you can As many as you can afford.

Inhuman output needs inhuman controls. AI lets a small team ship more than anyone can check by hand. The controls have to be inhuman too. Extreme observability. Agents verifying each other. A harness that repairs itself +++

This is the right mindset for creatives when thinking about AI Congratulations 🎉 You have all been promoted to Creative Directors

internet in 2026 is very much alive and well

INVESTIGATE ALEX PADILLA

we now manage by prompt

it’s alive

everyone was talking about "loop engineering" about 6 months too early

insane value here 🔥

liverecover is now VYG Conversations. it started as cart recovery. now it's programmatic conversations for any business: send a conversation by api, a real person handles it. next: human decisions. vyg.ai/conversations

coming soon to vyg.ai

Companies should pay for their employees’ personal AI usage. That is, if you want them to be any good. @geoffreyhuntley has a great analogy - it’s like learning to play an instrument. You have to practice A LOT to make anything good. AI is the same...you need to let your team play with the latest models on inconsequential work to know their strengths and limitations. Rebuild your budget spreadsheet. Sort your photos. Plan a trip. Doesn’t matter. That’s where you learn when the model is brilliant and when it’s lying to you. You don’t want them to find out on production or with your customers. I also think about the classic mega corporation. All of the executives have an ensuite bathroom, dry cleaning service, travel agent +++ None of that was about the perks. It was buying back attention for the decisions that mattered. Now, EVERYONE is Daddy Warbucks, commanding a fleet of 100s+ workers moving at light speed. You WANT your team to be fully unlocked so they have mind space to make highly consequential decisions at light speed. Personal AI is the new ensuite bathroom. And it’s cheaper than the dry cleaning. One more thing. If the company pays and then owns everything you make with it, nobody builds anything real. People need a space that’s actually theirs. To toot our own horn...this is a foundational part of HQ. Each HQ user has their own identity, and with that a personal encrypted vault for projects, knowledge, workers, policies, data +++ Then, each company has its own encrypted vault, and individuals are granted access to that vault with set permissions. 100s of options exist for creating a “company brain”. But more valuable than that, and what HQ provides, is a framework for the separation of personal work from company work, with the underlying assumption that each individual has something to bring to the table. That they’re more than a cog in a company. When I first built HQ, I didn’t even envision all of the cloud, sync, and team features. I thought of it as a personal city. Each individual builds their own AGI, filled with their knowledge, workflows, skills and experience, competing and collaborating with others’ cities. Now HQ has the best of both worlds. Company work builds in a collective, compounding with each piece of work. Personal skills and processes compound around that, enabling creative chaos to do its magic. Everything compounds. Including the gap between teams that practice and teams that don’t. Pay for the practice room.

seriously... this is the most insane dashboard in eCommerce/DTC ever 100% free cdp and data hub (what GA used to be but way better) the UI is sick, but can connect MCP to claude and never open app again this is the future of commerce come help me design it 🤝 sign up @ vyg.ai now

This is AGI if you try it

What if every thing you did compounded? every conversation, design project, codebase, knowledge set, workflow... all in an eternal vault. this is every HQ user. start compounding.

AI doesn’t have a technology problem. It has a permission problem. My AI handles email, content, projects, team updates, and support tickets. It works because I own every function it touches. Now try that at a normal company. Marketing wants AI to rewrite the landing page. Engineering owns the page. Sales wants AI to use customer data. The data team owns the data. Every layer in an org chart is a permission boundary. AI that stops and waits for approval is just software with extra steps. And when it makes a mistake across a department line? Meetings. Lost trust. “Never let it touch our stuff again.” A 3-person startup where everyone does everything will get more out of AI than a 10,000-person company with an AI strategy. If changing a headline takes three approvals, AI won’t save you. The bottleneck was never the model. It was always the meeting.

should you use HQ? the choice is up to you. but prepare for ontological shock.

Opus 5.5 (fast) is so delightful i'm tempted to start paying for all of my tokens