
If you’re working on making computers easier in some capacity, please get in touch with me.

Engineer in Residence @SHV Working on new interfaces, tools, and ways to make computers better Sometimes investing in people doing the same

If you’re working on making computers easier in some capacity, please get in touch with me.

Also shout-out to @henrymodis for etching the phrase "garage project" into my brain

Sharing a garage project with you all soon - very excited to see what you all think of it!

we've partnered with @Microsoft to ship our neuron-derived ai video model locally on Windows 11 and powered by Windows ML. through our new TBC maker app, we use the speed of mischief to offer a complete ai video production experience, from idea to a multi-minute film, right on your own device.

you can see everything (your agents did). why would I deceive you? new in Intent mobile today: - view PRs in the Changes view - clearer reconnecting with a status pill - drag tools into the main area on iPad - rename terminals - faster with large diffs & workspaces - plus fixes & polish in release notes: intentapp.dev/updates

Regardless of how good AI’s are at writing code, having a deep understanding of things like computer science (as well as things like electrical engineering, materials science etc etc) will still be important for pushing the frontier
Computer Science is the most important thing you can study right now, but with an understanding of the centrality of language, semiotics, philosophy, economics, and biology. Skate to where the puck is going. Be positioned to shepherd people over to the new paradigm.

This is a difficult moment. It’s hard not to look at the journey this company has taken and not be filled with respect for what Jensen built. The spirit of the company remains strong: "mission is the boss”. That said, as NVIDIA finds its way forward, I want one thing to be clear: NVIDIA's contributions to open source are unwavering. Jensen has been vocal about this, and that commitment runs deep in NVIDIA's people and culture. We all understand the importance of open source. NVIDIA is a huge supporter of open source because we understand something. If you want to diffuse technology, you do so through an ecosystem, not a single company or entity. No one owns the internet, no one owns servers, and no one owns operating systems… but through open source, an ecosystem of closed and open innovation flourishes. When we talk about delivering agents safely and securely, that too will take place because of open source. For example, ensuring people have the best models to secure themselves at machine speed unburdened by arbitrary guardrails decided behind closed doors. Open source & AI are often discussed together, but it’s worth reminding, open source predates AI. It is a mechanism for sharing ideas inspired by the hippies at computer labs of California in the 1960s. Before AI, the largest shift I lived through was the internet. It was remarkable seeing the fight for open source take place there. Between browsers, runtimes, net neutrality, etc. What came as a result was HTML5 and an extremely vibrant ecosystem that built the internet as we know it today. NVIDIA and its partners hold an incredibly important role in delivering this technology. NVIDIA contributes to every layer of the stack. Runtimes, models, harnesses, programming languages, frameworks, etc. None of that changes. Long live open source.
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter. AI will transform every industry, power every company, and be built by every country. Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty. The world needs both frontier closed models and frontier open models. images.nvidia.com/pdf/Open-W…

new in Intent today: - redesigned workspace creation - GitLab support - use Cursor as agent provider - agent plugins for all providers - Product Manager agent - Maintenance Coordinator - pin agents & hide repos - usage charts by cache, input, output & thinking - plus more in release notes: intentapp.dev/updates

I was thinking about how you can ask an AI to try a reply again, but sometimes you like the beginning and just want to see where else it could’ve gone. I made a prototype, From Here, so you can put your cursor anywhere in a reply and let GPT-6 Luna take it from there. Everything before the cursor stays. You can also change a few words first and see what happens. You’ll need an OpenAI API key to try it: from-here.narner.chatgpt.sit…

I was thinking about how you can watch an AI reply stream in, but you never see what else could’ve come next. I made Other Words so you can click a token and see what else GPT-6 Luna might’ve picked. The percentages come from the model’s logprobs. Pick a different token and see where the sentence goes. You’ll need an OpenAI API key to try it: other-words.narner.chatgpt.s…


Speed is a feature
One of the things we should do with AI agents is put them on making all software much much faster and more responsive. It's very measurable, they should be able to iterate on it independently and optimize!

Mecha has a website now. Check it out at mechacorp.ai Reach out if you’re interested in building the robots that will be deployed to cities, businesses, and homes.

Wondering this morning what the last time I manually did a rebase or resolve a merge conflict was

Big thanks to @csvoss for bouncing ideas and feedback!

Tyler is right 10/10 no notes
Quick rant and thoughts on the state of software design & engineering. It comes as no surprise that the models have allowed designers to implement and explore their wildest dreams. But I cannot explain enough how much more important that makes understanding the lowest levels of the stack Take for example these demos. It turns out that instant launch, a fast + instantly zoomable grid, and morphing photo clusters onto the surface of a globe are actually all underpinned by the same technical component: a really fast image thumbnail cache (and cache management). There is a lot more detail that goes into making each experience independently high quality, but that cache is the biggest performance bottleneck for each one - because ultimately they all depend on extremely fast data retrieval to even exist. You can’t use a globe that doesn’t spin and you can’t use a grid you can’t see or scroll. Therefore they never get built and die in imagination land. Achieving your design goals requires being able to identify their core infrastructural overlap Obviously you don’t need to know every detail, but it’s now your responsibility now to use the models to help you uncover your unknown unknowns about the hidden connections in any system you are building. The proper solution is latent in your unoptimized code. Its complexity is the raw material you need to chisel away at and navigate to the “correct” side of the solution space Will the models be able to come to conclusions like this autonomously soon? Yes. I think there will need to be advancements in the ability to quickly reason about and quantify the complexity (whatever that means) of a given piece of software and instantly determine what to chop up / refactor / redesign as it develops into its simplest possible form. That might not mean the fewest or tersest LOC but perhaps how few turns it takes a model to achieve a verified solution to any given task, or how tangled its code reference graph is. But I think it will always be an advantage to know where to poke