id stringlengths 19 88 | platform stringclasses 1
value | url stringlengths 36 105 | published_at stringlengths 24 24 | title stringlengths 3 72 | text stringlengths 59 6.24k | link stringclasses 25
values | media listlengths 0 12 | tags listlengths 0 5 |
|---|---|---|---|---|---|---|---|---|
site:2025/Oct/9/hello-world | site | https://andysmith.ai/2025/Oct/9/hello-world/ | 2025-10-09T17:09:00.000Z | Hello World | Hi there.
My name is Andy and this is my blog.
To begin I would like to write some points about this.
1. This is a blog. All of these posts are my personal opinions. Not financial advice. Not an absolute truth. Just my opinion. Don't read it if you don't like it.
2. I write for myself in general. I need a place for ... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2025/Oct/9/hello-world/featured.png",
"text": null
}
] | [
"writing",
"philosophy",
"ai-agents",
"digital-identity"
] |
site:2025/Oct/10/why-ai-is-the-most-important-thing-to-invest-in-today | site | https://andysmith.ai/2025/Oct/10/why-ai-is-the-most-important-thing-to-invest-in-today/ | 2025-10-10T19:58:00.000Z | Why AI is the most important thing to invest in today | Let me explain why I think AI is the most important thing I should invest my time and money in.
I mean I should read news about it immediately after I wake up, before any other things. I should invest my time in reviewing new technologies, new models, and new architectural approaches. I should build my new company bas... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2025/Oct/10/why-ai-is-the-most-important-thing-to-invest-in-today/featured.png",
"text": null
}
] | [
"ai-adoption",
"disruption",
"entrepreneurship",
"future-of-work"
] |
site:2025/Oct/12/why-i-write-posts-myself-and-don-t-use-ai-generation | site | https://andysmith.ai/2025/Oct/12/why-i-write-posts-myself-and-don-t-use-ai-generation/ | 2025-10-12T15:42:00.000Z | Why I write posts myself and don't use AI generation | It can be unexpected to write texts about AI without AI. Let me explain why this is important for me.
First of all, this is not a magazine or a newspaper. These are just my essays about our futuristic days. I don't need to be perfect for anyone else, so I can make mistakes, fix them, study and become better right here... | null | [] | [
"writing",
"ai-writing",
"blogging",
"learning"
] |
site:2025/Oct/13/everything-should-be-written | site | https://andysmith.ai/2025/Oct/13/everything-should-be-written/ | 2025-10-13T14:49:00.000Z | Everything should be written | I'm building a habit of writing down everything. I think this is the base of efficient work.
Any thought should be written down. Any bookmark, any interesting book, any link, any quote, any paper should be saved. Everything interesting. But I should save only things which I am really interested in, not all of them. Th... | null | [] | [
"note-taking",
"personal-knowledge-management",
"writing",
"blogging"
] |
site:2025/Oct/17/organize-and-delegate-not-just-replace-yourself | site | https://andysmith.ai/2025/Oct/17/organize-and-delegate-not-just-replace-yourself/ | 2025-10-17T18:39:00.000Z | Organize and delegate, not just replace yourself | I watch in horror as people around me use AI in an absolutely unsafe way.
People give full access to their devices. The AI tools can use browsers (with all accounts), SMS, emails, bank apps. Once I saw somebody allow AI to attach by SSH to a production server! And nobody sees the problem with this.
What is the proble... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2025/Oct/17/organize-and-delegate-not-just-replace-yourself/featured.png",
"text": null
}
] | [
"ai-agent",
"security",
"privacy",
"least-privilege",
"delegation"
] |
site:2025/Oct/18/the-last-human-bastion-is-agency | site | https://andysmith.ai/2025/Oct/18/the-last-human-bastion-is-agency/ | 2025-10-18T11:14:00.000Z | The last human bastion is agency | Agency (in the context of making decisions) is the most important skill during the AI era.
I believe computers (robots/AI/agents) can do almost all human work. We can automate every task but not decisions.
Making decisions requires willpower (see Baumeister et. al.), but only humans have it.
Computers can **help** ... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2025/Oct/18/the-last-human-bastion-is-agency/image.jpeg",
"text": null
}
] | [
"agency",
"decision-making",
"automation",
"willpower",
"entrepreneurship"
] |
site:2026/Jan/10/command-controlling-the-world-through-text | site | https://andysmith.ai/2026/Jan/10/command-controlling-the-world-through-text/ | 2026-01-10T14:12:00.000Z | Command: Controlling the World Through Text | Command — controlling the "world" through text.
The idea is to work in a single interface — whether Obsidian, Commacs (a commander based on Emacs), or a custom-built one. The implementation doesn't matter; what matters is the technology itself.
Here's how it works: I make a "move" by writing a message (text, voice, i... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/10/command-controlling-the-world-through-text/image.png",
"text": null
}
] | [
"ai-agents",
"claude-code",
"automation",
"git"
] |
site:2026/Jan/14/better-written-than-lost | site | https://andysmith.ai/2026/Jan/14/better-written-than-lost/ | 2026-01-14T07:03:00.000Z | Better Written Than Lost | Originally, I planned to write all my texts myself without using AI. But now I realize this requires considerable—actually, serious—discipline and extra time to polish my style in a non-native language. So I often simply don't find the time, and that's why I don't write on my blog. Not because I have nothing to say.
S... | null | [] | [
"writing",
"blogging",
"ai-writing",
"automation"
] |
site:2026/Jan/14/done-today-beats-perfect-never | site | https://andysmith.ai/2026/Jan/14/done-today-beats-perfect-never/ | 2026-01-14T07:35:00.000Z | Done Today Beats Perfect Never | So I'm simplifying the writing process to the maximum, shortening the path from forming an idea in my head to publishing it.
First, I've been using a private Obsidian for personal notes for a long time. In one of my next posts, I'll describe it in more detail (along with the history of updating my note-taking systems,... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/14/done-today-beats-perfect-never/featured.png",
"text": null
}
] | [
"writing",
"publishing",
"automation",
"claude-code",
"obsidian"
] |
site:2026/Jan/14/everything-is-text | site | https://andysmith.ai/2026/Jan/14/everything-is-text/ | 2026-01-14T07:57:00.000Z | Everything Is Text | Reading [Cursor's post on dynamic context discovery](https://cursor.com/blog/dynamic-context-discovery).
"Treat terminal as a file" — that's the Emacs way. Both approaches share the same idea: a unified interface to everything through files (or buffers).
It would be fascinating to build an AI agent that works nativel... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/14/everything-is-text/featured.png",
"text": null
}
] | [
"emacs",
"ai-agent",
"cursor",
"plain-text"
] |
site:2026/Jan/15/unrealized-risks | site | https://andysmith.ai/2026/Jan/15/unrealized-risks/ | 2026-01-15T07:14:00.000Z | Unrealized Risks Are Still Real Costs | Reading "Fooled by Randomness" by Nassim Taleb.
Taking lots of notes as I go — I'll sort through them when I finish and maybe write something more detailed.
But right now one simple thought keeps spinning in my head (unsurprisingly, since it runs as a thick thread through the entire book). It's long been obvious to m... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/15/unrealized-risks/featured.png",
"text": null
}
] | [
"risk",
"decision-making",
"probability",
"expected-value",
"nassim-taleb"
] |
site:2026/Jan/15/my-digital-twin-starts-with-claude-code | site | https://andysmith.ai/2026/Jan/15/my-digital-twin-starts-with-claude-code/ | 2026-01-15T16:23:11.000Z | My Digital Twin Starts with Claude Code | I want to save everything I write in Claude Code. And everything it answers. Everything it does, which tools it calls. Absolutely everything. Then I want to index this information.
After that, I'll analyze it. What did I think about most this week? Where do I make mistakes most often? How can I optimize my work? What ... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/15/my-digital-twin-starts-with-claude-code/featured.png",
"text": null
}
] | [
"claude-code",
"digital-twin",
"personal-knowledge-graph",
"quantified-self",
"automation"
] |
site:2026/Jan/15/constraints-drive-workflows | site | https://andysmith.ai/2026/Jan/15/constraints-drive-workflows/ | 2026-01-15T17:14:22.000Z | No Way Back: Why Constraints Drive Better Workflows | I never run code locally. It's not safe. Instead, I always use sandboxes. Once upon a time it was a remote machine with Emacs that I connected to via SSH. Then I moved to GitHub Codespaces with VS Code. Then it became a remote machine again, this time with [VS Code tunnels](https://code.visualstudio.com/docs/remote/tun... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/15/constraints-drive-workflows/featured.png",
"text": null
}
] | [
"claude-code",
"developer-workflow",
"ai-agents",
"remote-development",
"sandboxing"
] |
site:2026/Jan/20/how-many-tokens-on-this-task | site | https://andysmith.ai/2026/Jan/20/how-many-tokens-on-this-task/ | 2026-01-20T08:15:28.000Z | How Many Tokens Did You Spend on This Task? | "How many tokens did you spend on this task?"
This is the question I ask when someone starts telling me that a task is complex, unsolvable, or that they don't understand how to approach it.
Of course, you can't get an exact answer to this question. Unless you're using my LLM accounting and observability system, but t... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/20/how-many-tokens-on-this-task/featured.png",
"text": null
}
] | [
"llm",
"tokens",
"observability",
"claude"
] |
site:2026/Jan/20/robots-should-do-everything | site | https://andysmith.ai/2026/Jan/20/robots-should-do-everything/ | 2026-01-20T10:30:00.000Z | Robots Should Do Everything | My entrepreneurial hypothesis for the coming year:
Robots should do everything.
Absolutely everything. I should only monitor how they do it and adjust when necessary.
I should build systems of robots.
My task comes down to creating and maintaining infrastructure so that robots can do their work.
Here we should con... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/20/robots-should-do-everything/image.png",
"text": null
}
] | [
"ai-agent",
"automation",
"multi-agent-system",
"entrepreneurship"
] |
site:2026/Jan/21/from-solo-sessions-to-agent-orchestras | site | https://andysmith.ai/2026/Jan/21/from-solo-sessions-to-agent-orchestras/ | 2026-01-21T07:37:45.000Z | From Solo Sessions to Agent Orchestras | The tactic is simple. Figure out how to consume all available compute, then find ways to increase that availability.
I have a Claude Code subscription and other AI tools. My goal is to use them as efficiently as possible. This means pushing token usage closer to 100% of the available limit while maintaining acceptable... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/21/from-solo-sessions-to-agent-orchestras/image.png",
"text": null
}
] | [
"claude-code",
"ai-agents",
"agent-orchestration",
"automation",
"compute-utilization"
] |
site:2026/Jan/22/capsules-isolated-environments-for-ai-agents | site | https://andysmith.ai/2026/Jan/22/capsules-isolated-environments-for-ai-agents/ | 2026-01-22T10:36:04.000Z | Capsules: Isolated Environments for AI Agents | My Evolution of Working with AI Tools. The Capsule Concept.
I use LLMs and coding agents extensively in my work. This post contains a brief history of my observations and hypotheses.
The first decision I made: never run agents locally. This is a rule for me. There are two reasons for this. First, I'm afraid the agent... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/22/capsules-isolated-environments-for-ai-agents/image.png",
"text": null
}
] | [
"ai-agents",
"sandboxing",
"nix",
"claude-code"
] |
site:2026/Jan/27/product-over-technology | site | https://andysmith.ai/2026/Jan/27/product-over-technology/ | 2026-01-27T10:43:54.000Z | Product Over Technology | I've spent my entire life working with technology. Studying it, building it, growing it. When evaluating any project, I instinctively reach for the technology lens first. How well is the code written? Does the architecture allow for future growth? What frameworks are being used?
When developing my own projects, I alwa... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/27/product-over-technology/image.png",
"text": null
}
] | [
"product",
"entrepreneurship",
"engineering-mindset"
] |
site:2026/Jan/27/botlord-human-as-corporation | site | https://andysmith.ai/2026/Jan/27/botlord-human-as-corporation/ | 2026-01-27T13:36:15.000Z | Botlord: The Human as Corporation | I've always tried to see myself as a company. Even as an employee, I viewed my employer as a client or partner. The problem is, I did it poorly. I only recently realized that I think in processes rather than product-outcomes, and that needs to change (see: [Product Over Technology](https://andysmith.ai/blog/2026/01/pro... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/27/botlord-human-as-corporation/image.png",
"text": null
}
] | [
"ai-agents",
"management",
"entrepreneurship",
"future-of-work"
] |
site:2026/Jan/28/clawdbot-capsules-and-self-evolving-agents | site | https://andysmith.ai/2026/Jan/28/clawdbot-capsules-and-self-evolving-agents/ | 2026-01-28T08:07:08.000Z | Clawdbot, Capsules, and Self-Evolving Agents | Yesterday all channels were buzzing about Clawdbot. I decided to install it and give it a try.
I liked the concept. Following my thinking from [Product Over Technology](https://andysmith.ai/blog/2026/01/product-over-technology/), I always consider the product concept and how to sell it first. The execution is not grea... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/28/clawdbot-capsules-and-self-evolving-agents/image.png",
"text": null
}
] | [
"ai-agents",
"clawdbot",
"capsules",
"product-strategy"
] |
site:2026/Jan/29/small-steps-beat-big-leaps-familiar-interfaces-win | site | https://andysmith.ai/2026/Jan/29/small-steps-beat-big-leaps-familiar-interfaces-win/ | 2026-01-29T15:22:10.000Z | Small Steps Beat Big Leaps: Why Familiar Interfaces Win | My hypothesis is simple. The recipe for success is to offer something simple and familiar (lowering the barrier to entry) and add value on top of it. Products that work on interfaces users already understand take off. Products that completely change how people work don't.
Consider these examples.
Cursor is VS Code, f... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/29/small-steps-beat-big-leaps-familiar-interfaces-win/image.png",
"text": null
}
] | [
"product-design",
"user-adoption",
"developer-experience",
"cursor",
"claude-code"
] |
site:2026/Jan/29/autonomy-with-acceptable-quality | site | https://andysmith.ai/2026/Jan/29/autonomy-with-acceptable-quality/ | 2026-01-29T15:49:45.000Z | The Real Metric for Autonomous AI: Quality Without Hand-Holding | The primary metric I use to evaluate my autonomous AI systems: autonomy of work with acceptable quality of result. This seems obvious from the name "autonomous," but until you articulate it explicitly, it's not.
I see people around me setting up multiple monitors to watch several parallel Claude Code sessions simultan... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/29/autonomy-with-acceptable-quality/image.png",
"text": null
}
] | [
"ai-agents",
"autonomy",
"agent-orchestration",
"claude-code"
] |
site:2026/Jan/30/short-sessions-beat-long-ones-in-claude-code | site | https://andysmith.ai/2026/Jan/30/short-sessions-beat-long-ones-in-claude-code/ | 2026-01-30T07:26:16.000Z | Short Sessions Beat Long Ones in Claude Code | When actively using Claude Code in manual mode, I always apply the same technique. I keep sessions as short as possible. I set a task, get the result, close the chat. If I realize I need to return to the task to clarify something, I do so via `claude -r`.
This workflow creates a need to save information between contex... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/30/short-sessions-beat-long-ones-in-claude-code/image.png",
"text": null
}
] | [
"claude-code",
"context-management",
"ai-agents",
"developer-workflow"
] |
site:2026/Jan/30/ai-wont-teach-you-unless-you-make-it | site | https://andysmith.ai/2026/Jan/30/ai-wont-teach-you-unless-you-make-it/ | 2026-01-30T09:48:12.000Z | AI Won't Teach You — Unless You Make It | I've been reading Anthropic's research on the [problems with AI assistance when acquiring new skills](https://www.anthropic.com/research/AI-assistance-coding-skills) ([arXiv paper](https://arxiv.org/abs/2601.20245)).
The hypothesis: AI accelerates your work where you're already an expert but hinders you from becoming ... | null | [
{
"type": "image",
"url": "https://andysmith.ai/2026/Jan/30/ai-wont-teach-you-unless-you-make-it/image.png",
"text": null
}
] | [
"ai-assisted-learning",
"ai-agent",
"automation",
"anthropic"
] |
site:2026/May/25/lazy-mcp-dispatch-and-discovery | site | https://andysmith.ai/2026/May/25/lazy-mcp-dispatch-and-discovery/ | 2026-05-25T00:00:00.000Z | Lazy MCP: Dispatch and Discovery | MCP is now used everywhere, and some products exist only as MCP servers. Some of these MCPs contain dozens or even hundreds of tools, and that creates a real problem when working with them.
## The problem
The manifest of such an MCP is shipped in every LLM call, which means token usage grows proportionally with the s... | null | [] | [
"mcp",
"ai-agents",
"function-calling",
"context-engineering"
] |
site:2026/Aug/11/a-blog-for-author-based-writing | site | https://andysmith.ai/2026/Aug/11/a-blog-for-author-based-writing/ | 2026-08-10T17:35:12.000Z | A blog for author-based writing | I spent a few hours today and yesterday working out a strategy for social media and being public.
I want social media to be a nice storefront, showing only the good trail. But what I write is raw thoughts, from the angles I actually care about.
I ended up landing on a split between "author-based" writing and "reader-... | null | [] | [
"writing",
"publishing",
"social-media",
"discourse"
] |
site:2026/Aug/11/blogging-in-the-age-of-ai | site | https://andysmith.ai/2026/Aug/11/blogging-in-the-age-of-ai/ | 2026-08-11T09:06:56.000Z | Blogging in the age of AI | In the age of AI, it's really tempting to automate blogging, build content factories, all of that.
But you have to split this by what you're writing for.
Marketing copy you can and should write with an LLM. The LLM makes it better, tunes it to the right audience, checks for mistakes, and so on.
But the stuff you wri... | null | [] | [
"writing",
"blogging",
"ai-agents",
"automation"
] |
site:2026/Aug/11/speech-to-text | site | https://andysmith.ai/2026/Aug/11/speech-to-text/ | 2026-08-11T09:49:31.000Z | Speech to text | I figured I'd look into how speech-to-text is doing these days.
Locally, I use [Handy](https://handy.computer) for input.

I tried comparing a few tools, like [openwhispr](https://github.com/OpenWhispr/openwhispr) an... | null | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/5ec3834eca31218749a99e2c31ba8119400dea92/speech-to-text.webp",
"text": "image"
}
] | [
"speech-to-text",
"local-first",
"open-source",
"handy"
] |
site:2026/Aug/11/a-speech-recognition-tool | site | https://andysmith.ai/2026/Aug/11/a-speech-recognition-tool/ | 2026-08-11T13:36:05.000Z | A speech recognition tool | I needed to transcribe a call between two people.
I looked for something ready-made and didn't find anything that fit. I wanted it to just work in one click on Apple MLX, split up who said what, and run LLM post-processing on top (to strip out the mumbling and filler words).
I decided to throw together a small Python... | null | [] | [
"speech-to-text",
"whisper",
"speaker-diarization",
"mlx"
] |
site:2026/Aug/11/tags-on-my-blog | site | https://andysmith.ai/2026/Aug/11/tags-on-my-blog/ | 2026-08-11T16:46:51.000Z | Tags on my blog | What to do about tags on a blog.
The point of tags is to highlight the topics I keep coming back to.
One problem with tags is that they're unstable. They depend on how deep you've gone into a topic.
At first, when I'm just starting to dig into a new topic, say AI, every post gets the same high-level tag: AI. Then as... | null | [] | [
"tagging",
"blogging",
"automation",
"taxonomy"
] |
site:2026/Aug/12/posthog-monitoring-through-the-user-s-eyes | site | https://andysmith.ai/2026/Aug/12/posthog-monitoring-through-the-user-s-eyes/ | 2026-08-12T11:31:56.000Z | PostHog: monitoring through the user's eyes | I've used PostHog for years as an open source alternative to Google Analytics. Today I read the docs and realized it's a deeper tool that flips the whole approach to infrastructure monitoring.
Traditionally the technical data about the state of a system (logs, metrics, traces) lives in a separate engineering system (C... | null | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/c93236cdcf7e2816a3f99981ff92360db10085cb/posthog-monitoring-through-the-user-s-eyes.webp",
"text": "image"
}
] | [
"posthog",
"observability",
"monitoring",
"product-analytics"
] |
site:2026/Aug/13/is-there-any-demand-for-owning-your-own-data | site | https://andysmith.ai/2026/Aug/13/is-there-any-demand-for-owning-your-own-data/ | 2026-08-13T05:14:57.000Z | Is there any demand for owning your own data? | The classic SaaS business model is this: the user hands their data over to a provider, and in exchange they get an easy onboarding.
The risks are obvious. The user doesn't know how the provider handles their data, and can't do anything about the risk of leaks. The provider can also lock the account without warning, or... | null | [] | [
"data-ownership",
"self-hosting",
"open-source",
"ai-agents",
"saas"
] |
site:2026/Aug/13/training-a-personal-llm-on-what-you-actually-do | site | https://andysmith.ai/2026/Aug/13/training-a-personal-llm-on-what-you-actually-do/ | 2026-08-13T13:06:45.000Z | Training a personal LLM on what you actually do | I'm reading about Gwern's Guardian Angels (https://gwern.net/guardian-angel).
The idea is that instead of using general LLMs (which are trained to be useful to everyone, hackers included), you train your own personal LLM.
Instead of putting some bit of info about your personality into the context (which is never comp... | null | [] | [
"personal-llm",
"fine-tuning",
"gwern",
"emacs",
"digital-identity"
] |
site:2026/Aug/14/a-personal-public-edu-stream | site | https://andysmith.ai/2026/Aug/14/a-personal-public-edu-stream/ | 2026-08-14T05:38:53.000Z | A personal public EDU stream | Reading tons of text from AI agents wears my eyes out, so I've been thinking about other ways to interact with them.
The obvious one is voice.
The simplest version is just a voice assistant that answers questions, but that's not interesting. The default ChatGPT apps already do it.
Taking the idea further, I landed o... | null | [] | [
"ai-agents",
"voice-interface",
"discord",
"personalization"
] |
site:2026/Aug/14/a-stream-of-reactions-as-a-user-profile | site | https://andysmith.ai/2026/Aug/14/a-stream-of-reactions-as-a-user-profile/ | 2026-08-14T05:50:43.000Z | A stream of reactions as a user profile | Thought through an idea from https://andysmith.ai/2026/Aug/13/training-a-personal-llm-on-what-you-actually-do/ a bit more.
The thing you store and process isn't an "event" or an "active action." It's a reaction to some stimulus. The stimulus here is context, a description of the state of the world around the user. The... | null | [] | [
"data-ownership",
"blockchain",
"zk-rollup",
"llm",
"personalization"
] |
site:2026/Aug/15/looking-at-markets-from-first-principles | site | https://andysmith.ai/2026/Aug/15/looking-at-markets-from-first-principles/ | 2026-08-15T07:20:50.000Z | Looking at markets from first principles | I look at markets from first principles: what systems are in there, what roles exist, and how those roles get pushed through their lifecycle.
From there, if I want to make money, I have to pick a role and play it well.
Can I create a new role? Or is "find" the better word here? The role already exists, and maybe some... | null | [] | [
"markets",
"first-principles",
"strategy",
"mental-models"
] |
site:2026/Aug/15/what-if-i-used-my-own-social-media-to-learn | site | https://andysmith.ai/2026/Aug/15/what-if-i-used-my-own-social-media-to-learn/ | 2026-08-15T07:24:10.000Z | What if I used my own social media to learn? | I write blog posts, off the top of my head, about whatever I'm thinking about.
Agents post to social media as me, but they generalize my writing up to some known model or theory.
Then they look at those theories through my own writing.
The point is that I'd actually want to read the results. My own ideas run through... | null | [] | [
"ai-agents",
"automation",
"writing",
"social-media"
] |
site:2026/Aug/15/infinite-context | site | https://andysmith.ai/2026/Aug/15/infinite-context/ | 2026-08-15T07:28:15.000Z | Infinite context | When you put together a history of active actions (https://andysmith.ai/2026/Aug/14/a-stream-of-reactions-as-a-user-profile/), the context is all of life.
Every reaction is the sum of everything that happened in my life up to the moment it happens.
How do you capture that to build a training set?
One option: attach ... | null | [] | [
"machine-learning",
"training-data",
"context"
] |
site:2026/Aug/15/setting-up-an-open-source-reading-stack | site | https://andysmith.ai/2026/Aug/15/setting-up-an-open-source-reading-stack/ | 2026-08-15T10:15:46.000Z | Setting up an open-source reading stack | I try to read not just through agents, but the original sources too.
I want to read everything in one place: PDF papers, web pages, books.
I used to do all of this in Zotero (annotations too), but the problem is the server is closed and paid, and WebDAV plays badly with the API. I never managed to give agents proper ... | null | [] | [
"reading",
"self-hosting",
"open-source",
"zotero",
"bookorbit"
] |
site:2026/Aug/15/the-company-as-a-system | site | https://andysmith.ai/2026/Aug/15/the-company-as-a-system/ | 2026-08-15T10:45:44.000Z | The company as a system | Instead of forcing old models onto agent work (where everyone runs their own agents and automates their own piece), we should aim to build the company as a set of agents that run all the time and talk to each other.
In this setup, people (backed by their own personal agents) either build and keep the system running, o... | null | [] | [
"ai-agents",
"automation",
"organization",
"future-of-work"
] |
site:2026/Aug/16/a-universal-voice-assistant | site | https://andysmith.ai/2026/Aug/16/a-universal-voice-assistant/ | 2026-08-16T06:48:15.000Z | A universal voice assistant | I'm building on my idea for a personal EDU stream (https://andysmith.ai/2026/Aug/14/a-personal-public-edu-stream/).
The idea is a universal voice assistant where one of the modes is learning.
This assistant can also run all my sessions in Claude Code (and other agents), answer email, and do everything some Hermes-lik... | null | [] | [
"voice-assistant",
"speech-to-text",
"ai-agents",
"claude-code",
"learning"
] |
site:2026/Aug/16/context-as-a-separate-stream-of-events | site | https://andysmith.ai/2026/Aug/16/context-as-a-separate-stream-of-events/ | 2026-08-16T06:48:29.000Z | Context as a separate stream of events | One way to build context-reaction pairs, while keeping [infinite context](https://andysmith.ai/2026/Aug/15/infinite-context/) in mind, goes like this.
Call everything context: everything I see, hear, and feel. Everything around me is context. Split it into quanta (events) and write them to a separate append-only log. ... | null | [] | [
"llm",
"lifelogging",
"wearables",
"context-engineering"
] |
site:2026/Aug/16/why-devops-matters | site | https://andysmith.ai/2026/Aug/16/why-devops-matters/ | 2026-08-16T06:49:26.000Z | Why DevOps matters | I caught myself realizing I've stopped reading the code an LLM writes closely.
Instead, I put my attention on everything around it: tests, observability, quality metrics.
That makes DevOps far more important. You have to catch a problem as early as you can and keep it off prod. Tests have to work, always. Metrics and... | null | [] | [
"devops",
"observability",
"testing",
"llm"
] |
site:2026/Aug/16/a-kanban-board-for-my-ai-agents | site | https://andysmith.ai/2026/Aug/16/a-kanban-board-for-my-ai-agents/ | 2026-08-16T06:49:51.000Z | A kanban board for my AI agents | I've tried a lot of different AI interfaces: copilot, chats, console agents, bots in direct and group chats in messengers.
They all miss the same thing for me: async.
Right now I'm experimenting with kanban boards.
One board, four columns:
- Backlog: the task needs my attention. I take the first card, get into the ... | null | [] | [
"ai-agents",
"kanban",
"forgejo",
"automation"
] |
site:2026/Aug/16/a-context-reaction-tree-as-a-zk-proof-of-expertise | site | https://andysmith.ai/2026/Aug/16/a-context-reaction-tree-as-a-zk-proof-of-expertise/ | 2026-08-16T07:52:08.000Z | A context/reaction tree as a ZK proof of expertise | If you keep a careful tree of your context and reactions (https://andysmith.ai/2026/Aug/16/context-as-a-separate-stream-of-events/), you can use it as proof of competence.
It could even become a new kind of résumé.
Say I'm an expert in AI, and I also know how to build birdhouses. But for some reason I'd rather not ma... | null | [] | [
"zero-knowledge-proofs",
"privacy",
"reputation",
"personal-knowledge-management"
] |
site:2026/Aug/16/a-personal-ai-assistant-as-a-butler | site | https://andysmith.ai/2026/Aug/16/a-personal-ai-assistant-as-a-butler/ | 2026-08-16T09:40:28.000Z | A personal AI assistant as a butler | An AI agent like Hermes or OpenClaw can do real work out in the world, like writing code or filing bug reports. So it's tempting to use it for exactly that.
I see it differently.
To me, a personal assistant is an agent system that shouldn't do any of that outside work itself. Instead it orchestrates the other systems... | null | [] | [
"ai-agents",
"personal-assistant",
"local-first",
"orchestration"
] |
site:2026/Aug/16/auto-publishing-to-my-telegram-channel | site | https://andysmith.ai/2026/Aug/16/auto-publishing-to-my-telegram-channel/ | 2026-08-16T14:35:00.000Z | Auto-publishing to my Telegram channel | I set up automatic publishing of my posts to a Telegram channel.
I decided to post them there raw, with no AI reworking. I called it brain dump. Same as on the site, I'll put down whatever's left in my head once everything else is automated. And something always is. You can't just think about nothing.
Now I write any... | null | [] | [
"telegram",
"automation",
"publishing",
"blogging"
] |
site:2026/Aug/17/infrastructure-as-a-formal-system | site | https://andysmith.ai/2026/Aug/17/infrastructure-as-a-formal-system/ | 2026-08-17T07:45:50.000Z | Infrastructure as a formal system | I'm reading (well, listening to) Gödel, Escher, Bach by Douglas Hofstadter.
The chapter on formal systems caught my eye, because you might be able to apply it to infrastructure.
The goal: infrastructure that's guaranteed to be secure.
The idea is to describe infrastructure as a formal system where the theorems are t... | null | [] | [
"infrastructure",
"formal-methods",
"security",
"ai-agents"
] |
site:2026/Aug/17/the-shared-context-problem-on-teams-that-use-ai | site | https://andysmith.ai/2026/Aug/17/the-shared-context-problem-on-teams-that-use-ai/ | 2026-08-17T12:25:38.000Z | The shared context problem on teams that use AI | I keep running into this problem on small teams where everyone uses AI.
Person 1 writes code with their agents, which lean on person 1's context. They commit to the repo. Then person 2 makes changes, with agents that lean on person 2's context.
The two contexts are different, because each one can carry that person's ... | null | [] | [
"ai-agents",
"context-engineering",
"team-collaboration",
"multi-agent"
] |
site:2026/Aug/18/kanban-is-the-interface-between-a-human-and-an-agent-not-between-agents | site | https://andysmith.ai/2026/Aug/18/kanban-is-the-interface-between-a-human-and-an-agent-not-between-agents/ | 2026-08-18T09:59:45.000Z | Kanban is the interface between a human and an agent, not between agents | The kanban I'm working on is mainly for communication between a human and an agent. Between agents you don't need a kanban at all. That part can be whatever. But for a person, moving little cards around a board is just handy.
The idea is this. You start a session with an agent. It does its thing, and at some point it ... | null | [] | [
"ai-agents",
"kanban",
"human-in-the-loop",
"workflow"
] |
site:2026/Aug/18/collect-data-about-yourself-seamlessly-don-t-lean-on-discipline | site | https://andysmith.ai/2026/Aug/18/collect-data-about-yourself-seamlessly-don-t-lean-on-discipline/ | 2026-08-18T10:10:47.000Z | Collect data about yourself seamlessly, don't lean on discipline | I got talking with someone about collecting data on yourself. He was telling me how he logs the meals he eats through the claude.ai web interface and saves them all in one database.
I don't believe in collecting data on yourself through discipline. Discipline breaks. When you do something without really understanding ... | null | [] | [
"quantified-self",
"self-hosting",
"wearables",
"automation",
"apple-health"
] |
site:2026/Aug/19/an-experience-lake-and-keeping-personal-data-safe | site | https://andysmith.ai/2026/Aug/19/an-experience-lake-and-keeping-personal-data-safe/ | 2026-08-19T13:21:53.000Z | An experience lake, and keeping personal data safe | I'm thinking about storing the context of my whole life, every event and every reaction, as a graph. At first I wanted to call it a data lake. But experience lake is closer: everything you've built up over years.
The point of a lake like this is that you can open it up piece by piece. Say a company comes to you and sa... | null | [] | [
"privacy",
"data-ownership",
"encryption",
"knowledge-graph",
"ai-training"
] |
site:2026/Aug/19/a-personal-stream-with-voice-as-the-main-interface | site | https://andysmith.ai/2026/Aug/19/a-personal-stream-with-voice-as-the-main-interface/ | 2026-08-19T13:24:12.000Z | A personal stream, with voice as the main interface | I've been thinking for a while about what I call a personal stream, and now I finally have a chance to try it.
The idea is this. There's an endless feed, like on social media. That's the visual version. And there's a second version: you put in an earbud, and it's like radio. Smart radio that just reads the feed out to... | null | [] | [
"voice-interface",
"ai-agent",
"text-to-speech",
"personal-stream"
] |
site:2026/Aug/19/ai-doesn-t-work-in-a-team-and-how-to-fix-it | site | https://andysmith.ai/2026/Aug/19/ai-doesn-t-work-in-a-team-and-how-to-fix-it/ | 2026-08-19T13:27:13.000Z | AI doesn't work in a team, and how to fix it | To me, all these IDEs and Claudes seem to have one basic problem: they don't work as a team. You've got your own context, your own tricks, principles, gates, everything you've set up for yourself. My context is totally different. The devs have a third one. And when we're building one shared product, each of us playing ... | null | [] | [
"ai-agents",
"collaboration",
"coding-agents",
"developer-workflow"
] |
site:2026/Aug/19/self-description-and-a-permission-sandbox-for-agents-in-lisp | site | https://andysmith.ai/2026/Aug/19/self-description-and-a-permission-sandbox-for-agents-in-lisp/ | 2026-08-19T13:28:55.000Z | Self-description and a permission sandbox for agents in LISP | I've been listening to Gödel, Escher, Bach, and it feels like it's about exactly what I want to build: systems that describe themselves. You know those Escher drawings, the staircase that keeps going down but never ends, or the waterfall that pours down and somehow feeds itself back at the top. Those self-looping thing... | null | [] | [
"lisp",
"ai-agents",
"sandboxing",
"programming-languages"
] |
site:2026/Aug/20/a-programming-paradigm-for-spatiotemporal-composability | site | https://andysmith.ai/2026/Aug/20/a-programming-paradigm-for-spatiotemporal-composability/ | 2026-08-20T09:58:16.000Z | A Programming Paradigm for Spatiotemporal Composability | 
This paper uses math to explain concepts that are a lot like the ones I'm trying to build in Zeno.
But the authors use TypeScript, and that saddles them with a whole pile... | https://github.com/cordiverse/paper | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/23613c3577f3afdab8fff700bde09905d1197787/a-programming-paradigm-for-spatiotemporal-composability.webp",
"text": "screenshot"
}
] | [
"zeno",
"lisp",
"typescript",
"programming-languages"
] |
site:2026/Aug/21/the-open-source-llm-router-for-ai-agents-manifest | site | https://andysmith.ai/2026/Aug/21/the-open-source-llm-router-for-ai-agents-manifest/ | 2026-08-21T07:21:07.000Z | The Open Source LLM Router for AI Agents | Manifest | 
It's an LLM gateway, but with HA, fallback, and observability.
I want to try it out against LiteLLM. It'd be handy to funnel every agent's requests into one place, especially with t... | https://manifest.build/ | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/6a162b12c1f2b74a8a35caf2c31d1137d6646c65/the-open-source-llm-router-for-ai-agents-manifest.webp",
"text": "image"
}
] | [
"llm-gateway",
"litellm",
"ai-agents",
"observability",
"langfuse"
] |
site:2026/Aug/21/crossplane-is-the-cloud-native-framework-for-platform-engineering | site | https://andysmith.ai/2026/Aug/21/crossplane-is-the-cloud-native-framework-for-platform-engineering/ | 2026-08-21T07:35:39.000Z | Crossplane Is the Cloud-Native Framework for Platform Engineering | 
Interesting way to describe infrastructure, using the reconciliation loop from Kubernetes.
The idea is that you describe the desired state of your infra as a Kubernetes m... | https://www.crossplane.io/ | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/dfee53540d5d90714a44c37cf037a18c774998d9/crossplane-is-the-cloud-native-framework-for-platform-engine.webp",
"text": "image"
}
] | [
"crossplane",
"platform-engineering",
"infrastructure-as-code",
"kubernetes",
"pulumi"
] |
site:2026/Aug/21/openssf-best-practices-badge-program | site | https://andysmith.ai/2026/Aug/21/openssf-best-practices-badge-program/ | 2026-08-21T07:45:43.000Z | OpenSSF Best Practices Badge Program | 
A checklist and a badge for best practices in open source repos.
It's a lot like what I was trying to do with https://github.com/sm-th/show-your-work, but mine ended up being more about form and... | https://www.bestpractices.dev/ | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/e759541395626caab0da55433bc80b6d0820ff8c/openssf-best-practices-badge-program.webp",
"text": "image"
}
] | [
"open-source",
"best-practices",
"openssf",
"security"
] |
site:2026/Aug/21/omp-a-coding-agent-with-the-ide-wired-in | site | https://andysmith.ai/2026/Aug/21/omp-a-coding-agent-with-the-ide-wired-in/ | 2026-08-21T07:48:51.000Z | Omp -- a coding agent with the IDE wired in | 
I'm trying out OMP (oh-my-pi).
It's a batteries-included harness, and it works with your existing Claude/ChatGPT subscriptions.
From the description it's something like an IPython notebook ... | https://github.com/can1357/oh-my-pi | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/ea6aa7ade4ea7b44d11d40763c436fb2d0668afd/omp-a-coding-agent-with-the-ide-wired-in.webp",
"text": "image"
}
] | [
"omp",
"coding-agent",
"ai-agents",
"emacs",
"org-mode"
] |
site:2026/Aug/21/buzz-a-hive-mind-communication-platform | site | https://andysmith.ai/2026/Aug/21/buzz-a-hive-mind-communication-platform/ | 2026-08-21T10:55:22.000Z | Buzz: a hive mind communication platform | 
I've had my eye on this tool from Jack Dorsey for a while. Now I'm getting into it in more detail.
It's a messenger for people and agents. The obvious point: instead of making people get used... | https://github.com/block/buzz | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/f5cf53d3fbcd63c6242d74ec9846488286b68ee8/buzz-a-hive-mind-communication-platform.webp",
"text": "image"
}
] | [
"buzz",
"ai-agents",
"messaging",
"jack-dorsey"
] |
site:2026/Aug/21/nostr-notes-and-other-stuff-transmitted-by-relays | site | https://andysmith.ai/2026/Aug/21/nostr-notes-and-other-stuff-transmitted-by-relays/ | 2026-08-21T10:58:56.000Z | Nostr - Notes and Other Stuff Transmitted by Relays | 
Nostr is one tool for building a decentralized append-only log of events. That kind of log works well for decentralized social networks, messengers, and other systems built around st... | https://nostr.org/ | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/dfa68059d28ab072ec959624a68bb15d1e6d15c8/nostr-notes-and-other-stuff-transmitted-by-relays.webp",
"text": "image"
}
] | [
"nostr",
"decentralization",
"ai-agent",
"event-sourcing"
] |
site:2026/Aug/22/shottr-a-screenshot-app-for-pixel-professionals | site | https://andysmith.ai/2026/Aug/22/shottr-a-screenshot-app-for-pixel-professionals/ | 2026-08-22T09:45:05.000Z | Shottr, a screenshot app for pixel professionals | 
A handy tool for taking screenshots. You can grab them fast, annotate them, and save them straight to S3.
For my links blog I decided to take the screenshots by hand. I pick a useful ... | https://shottr.cc/ | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/8987ea8e9debf724b9e57f11b41dd03f8c6100c9/shottr-a-screenshot-app-for-pixel-professionals.webp",
"text": "image"
}
] | [
"shottr",
"screenshots",
"macos-apps",
"blogging"
] |
site:2026/Aug/23/an-org-mode-based-harness | site | https://andysmith.ai/2026/Aug/23/an-org-mode-based-harness/ | 2026-08-23T04:56:37.000Z | An org-mode based harness | The more I dig into [OMP](https://andysmith.ai/2026/Aug/21/omp-a-coding-agent-with-the-ide-wired-in/), the more it reminds me of Emacs org-mode (https://orgmode.org/), and specifically its org-babel module (https://orgmode.org/worg/org-contrib/babel/).
I want to try using org-mode files as an interface to agents.
I c... | null | [] | [
"org-mode",
"emacs",
"ai-agents",
"omp"
] |
site:2026/Aug/26/cloudflare-mesh | site | https://andysmith.ai/2026/Aug/26/cloudflare-mesh/ | 2026-08-26T01:38:59.000Z | Cloudflare Mesh | 
Cloudflare Mesh could be a decent replacement for Tailscale.
Cloudflare is already on almost every site anyway, so using it to protect your internal infrastructure makes sense.
But that raises the stakes too. If so... | https://developers.cloudflare.com/cloudflare-one/networks/connectors/cloudflare-mesh/ | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/8093d7e200c31810007f07046317390c45b0d016/cloudflare-mesh.webp",
"text": "image"
}
] | [
"cloudflare",
"tailscale",
"networking",
"self-hosting"
] |
site:2026/Aug/26/microsandbox-a-computer-for-anything-anywhere | site | https://andysmith.ai/2026/Aug/26/microsandbox-a-computer-for-anything-anywhere/ | 2026-08-26T11:11:50.000Z | Microsandbox, a computer for anything, anywhere | 
Microsandbox lets you run OCI images in a microVM. It works on Windows, Linux, and Apple Silicon, and you don't need Docker.
You can script the whole VM launch as code.
So for each age... | https://microsandbox.dev/ | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/ecbdbd67b67b01dbb70ef532252de21e61ca0262/microsandbox-a-computer-for-anything-anywhere.webp",
"text": "image"
}
] | [
"microsandbox",
"microvm",
"sandboxing",
"ai-agents",
"virtualization"
] |
site:2026/Aug/26/snapcompact-sota-compaction-instant-local-free-pick-3 | site | https://andysmith.ai/2026/Aug/26/snapcompact-sota-compaction-instant-local-free-pick-3/ | 2026-08-26T11:20:16.000Z | Snapcompact: SoTA compaction, instant, local, free. Pick 3 | 
A clever way to get context into a model while spending fewer tokens on it.
The idea: instead of the text, you hand the model an image with that same text written on it in a tin... | https://stencil.so/blog/snapcompact | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/1ab56843b4230eab76029f8f0301d5f2fa2969a7/snapcompact-sota-compaction-instant-local-free-pick-3.webp",
"text": "image"
}
] | [
"context-compaction",
"token-optimization",
"multimodal",
"snapcompact"
] |
site:2026/Aug/26/tauri-2-0 | site | https://andysmith.ai/2026/Aug/26/tauri-2-0/ | 2026-08-26T12:22:01.000Z | Tauri 2.0 | 
Tauri is a tool for turning web apps into desktop (and mobile) apps. It's an alternative to Electron.
The upside is size and speed. Tauri doesn't ship Chrome and a whole runtime with every app. It uses the system WebView ... | https://tauri.app/ | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/457fe89902665784622c19b48488fe40d71e67ae/tauri-2-0.webp",
"text": "image"
}
] | [
"tauri",
"electron",
"desktop-app",
"rust",
"web-development"
] |
site:2026/Aug/26/mesh-llm | site | https://andysmith.ai/2026/Aug/26/mesh-llm/ | 2026-08-26T15:30:47.000Z | Mesh LLM | 
A tool that links several machines into one network and shards a big LLM's inference across multiple devices.
[Buzz](https://andysmith.ai/2026/Aug/21/buzz-a-hive-mind-communication-platform/) uses it to build Buzz Mesh.
H... | https://meshllm.cloud/ | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/fb721e519e61ba4b4cfcec539400d09391088107/mesh-llm.webp",
"text": "image"
}
] | [
"distributed-inference",
"llm",
"self-hosting",
"buzz"
] |
site:2026/Aug/31/reconciling-documentation | site | https://andysmith.ai/2026/Aug/31/reconciling-documentation/ | 2026-08-31T12:38:13.000Z | Reconciling documentation | Instead of relying on chat context and separate frameworks for long-term memory, I'm testing a way of building software based on the reconciliation pattern from Kubernetes.
The idea is to read as little of the AI's output as possible while keeping the quality of the development process, and ideally improving it.
Here... | null | [] | [
"ai-agents",
"documentation",
"kubernetes",
"software-development",
"developer-workflow"
] |
site:2026/Aug/31/lisp-as-an-interface | site | https://andysmith.ai/2026/Aug/31/lisp-as-an-interface/ | 2026-08-31T12:38:23.000Z | Lisp as an interface | The idea: I write my agent's logic in Lisp, using primitives like (say "message") and (ask-y-n "question"). The interface for talking to the user, those same say and ask functions and maybe a few others, can be overridden by the user or shipped as plugins.
Some people like to work by voice, some through Telegram or Wh... | null | [] | [
"lisp",
"ai-agents",
"human-computer-interaction",
"llm"
] |
site:2026/Aug/31/discourse-as-a-platform-for-an-ai-native-company | site | https://andysmith.ai/2026/Aug/31/discourse-as-a-platform-for-an-ai-native-company/ | 2026-08-31T15:54:18.000Z | Discourse as a platform for an AI-native company | After trying out Buzz (https://andysmith.ai/2026/Aug/21/buzz-a-hive-mind-communication-platform/), I realized the idea is really cool, but the implementation is still a bit raw.
Buzz as a messenger just didn't click with me. Glitches pop up here and there.
Meanwhile I've had Discourse running for a long time. I write... | null | [] | [
"discourse",
"ai-agent",
"acp",
"mcp",
"team-communication"
] |
site:2026/Sep/1/rethinking-the-vision-for-reflection-castle | site | https://andysmith.ai/2026/Sep/1/rethinking-the-vision-for-reflection-castle/ | 2026-09-01T06:55:28.000Z | Rethinking the vision for Reflection Castle | I'm building [Reflection Castle](https://github.com/reflection-dev/castle) as "a home for agents and people (tower), plus infrastructure for communication and for storing code and docs (citadel)". All of it ran on NixOS or Nix-Darwin.
The plan is to ship an open-source solution you can drop into any team and have runn... | null | [] | [
"nixos",
"kubernetes",
"flux",
"ai-agents",
"self-hosting"
] |
site:2026/Sep/1/pijul | site | https://andysmith.ai/2026/Sep/1/pijul/ | 2026-09-01T07:10:19.000Z | Pijul | 
It's a version control system, an alternative to git.
I looked at it while hunting for somewhere to store the [experience lake](https://andysmith.ai/2026/Aug/16/context-as-a-separate-stream-of-events/), but it didn't fit.
St... | https://pijul.org/ | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/99aa2259f084409c4479c22ece58b02efcf2590c/pijul.webp",
"text": "image"
}
] | [
"pijul",
"version-control",
"git"
] |
site:2026/Sep/1/castle-architecture | site | https://andysmith.ai/2026/Sep/1/castle-architecture/ | 2026-09-01T10:52:40.000Z | Castle architecture | I'm still rethinking [Castle](https://andysmith.ai/2026/Sep/1/rethinking-the-vision-for-reflection-castle/).
I'm working from the idea that this could grow into a product you can easily deploy yourself. Or that I could walk into a company, set it up for them, and keep maintaining it afterward.
A three-layer architect... | null | [] | [
"castle",
"kubernetes",
"ai-agents",
"self-hosting"
] |
site:2026/Sep/1/onboarding-agents | site | https://andysmith.ai/2026/Sep/1/onboarding-agents/ | 2026-09-01T11:13:51.000Z | Onboarding agents | An addition to the Castle architecture: https://andysmith.ai/2026/Sep/1/castle-architecture/
The agent layer is a repo that manages agents. But agents need access, and each one needs different access.
So how do you hand that access out?
When I add a new agent, at a minimum I need to:
1. issue a token for the model
... | null | [] | [
"ai-agents",
"onboarding",
"access-control",
"automation",
"discourse"
] |
site:2026/Sep/2/a-devbox-on-a-remote-server | site | https://andysmith.ai/2026/Sep/2/a-devbox-on-a-remote-server/ | 2026-09-02T07:32:12.000Z | A devbox on a remote server | I really don't like running third-party software and agents on my main work machine. I can't always keep close track of what they're doing, and that makes me not trust them.
So I keep the software on my work laptop to the bare minimum. Everything else runs in isolated environments.
One thing I do is a remote devbox. ... | null | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/6127f9c14845dab360985ab67e683dfaf65bd210/a-devbox-on-a-remote-server.webp",
"text": "image"
}
] | [
"remote-development",
"vscode",
"hetzner",
"self-hosting",
"ai-agents"
] |
site:2026/Sep/2/bitnami-sealed-secrets | site | https://andysmith.ai/2026/Sep/2/bitnami-sealed-secrets/ | 2026-09-02T09:11:05.000Z | Bitnami/sealed-secrets | 
I use bitnami/sealed-secrets to deploy secrets to Kubernetes from sops.
That way I get one sops flow for both my NixOS secrets and my k8s secrets, all encrypted with age.
Lately though I've been looking at s... | https://github.com/bitnami/sealed-secrets | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/2e4af316bbdbc28ecb65aeb4923e95e675da1695/bitnami-sealed-secrets.webp",
"text": "image"
}
] | [
"sealed-secrets",
"sops",
"kubernetes",
"secrets-management",
"nixos"
] |
site:2026/Sep/2/bb-app | site | https://andysmith.ai/2026/Sep/2/bb-app/ | 2026-09-02T12:17:00.000Z | BB app | 
A harness that supposedly can modify itself.
Basically what I'm building in Zeno, but in NodeJS instead of Lisp.
It's fun to watch what comes out of projects like this (what use cases they cover) and then cover the same one... | https://getbb.app | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/d495b9ce3b959ab9c46b281b4119973da3a694fe/bb-app.webp",
"text": "image"
}
] | [
"ai-agents",
"self-modifying-code",
"zeno",
"harness"
] |
site:2026/Sep/2/lightpanda | site | https://andysmith.ai/2026/Sep/2/lightpanda/ | 2026-09-02T14:12:38.000Z | Lightpanda | 
I needed to give an agent a way to run a browser.
I'm trying out Lightpanda. The devs claim it's a lot faster than Chrome and uses a lot less memory. The final image is smaller too.
One problem: it can't take screenshot... | https://lightpanda.io | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/4881b7f39e463fbbff66e1bf55afe8a3424e70ca/lightpanda.webp",
"text": "image"
}
] | [
"lightpanda",
"browser-automation",
"ai-agent"
] |
site:2026/Sep/3/vllm | site | https://andysmith.ai/2026/Sep/3/vllm/ | 2026-09-03T12:59:48.000Z | vLLM | 
A toolkit for deploying LLM models to production.
It even does Apple MLX: https://github.com/vllm-project/vllm-metal | https://vllm.ai/ | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/eb2be7884c8352bf724a4a027f182889d6aa6f0c/vllm.webp",
"text": "image"
}
] | [
"vllm",
"llm",
"model-serving",
"mlx"
] |
site:2026/Sep/3/how-and-why-to-keep-agent-logs | site | https://andysmith.ai/2026/Sep/3/how-and-why-to-keep-agent-logs/ | 2026-09-03T13:01:43.000Z | How and why to keep agent logs | Once again I keep landing on the same thought: one way or another, the conversations with agents are the key artifact of this era.
From the logs you can pull out thinking patterns, mistakes, ideas. It's raw material for self-reflection. In the end, I believe the audit of the future is an audit of thinking, not of resu... | null | [] | [
"ai-agents",
"logging",
"observability",
"claude-code",
"self-hosting"
] |
site:2026/Sep/3/numtide-llm-agents | site | https://andysmith.ai/2026/Sep/3/numtide-llm-agents/ | 2026-09-03T13:41:22.000Z | Numtide/llm-agents | 
A huge, always-updated collection of LLM/AI tools.
Every tool ships with a Nix cache.
I use them as an omp cache, but I still drop by now and then to see what's new. | https://github.com/numtide/llm-agents.nix | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/117285de59d875021fc694a8a16c59ea9cf3b5c1/numtide-llm-agents.webp",
"text": "image"
}
] | [
"nix",
"llm-tools",
"ai-agents"
] |
site:2026/Sep/4/omarchy | site | https://andysmith.ai/2026/Sep/4/omarchy/ | 2026-09-04T06:19:01.000Z | Omarchy | 
Interesting project from the guy behind Ruby on Rails. A good-looking Linux. It's built for agents and run entirely by agents, which spooks me right off, given how agents love to go read your secrets. Not clear how they de... | https://omarchy.org | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/7bbe57e463d7dad696099bfaa103f8ab226509f6/omarchy-1.webp",
"text": "image"
},
{
"type": "image",
"url": "https://files.andysmith.ai/img/1b421e976eddafc0b623ec8cb143a3ab820c7a68/omarchy-2.webp",
"text": "IMG_9996"
}
] | [
"omarchy",
"linux",
"ai-agents",
"nixos"
] |
site:2026/Sep/4/16-9-screenshots | site | https://andysmith.ai/2026/Sep/4/16-9-screenshots/ | 2026-09-04T06:24:15.000Z | 16:9 screenshots | Here's how I make a Safari window 1280x720 (for some reason you have to set it a bit bigger, probably the Mac shadow or something like that).
```sh
osascript -e 'tell application "Safari" to set bounds of front window to {100, 100, 1380, 820}'
```
That gets me screenshots at exactly 16:9.
For terminal screenshots I ... | null | [] | [
"screenshots",
"wezterm",
"nix-darwin",
"macos"
] |
site:2026/Sep/4/hyprland | site | https://andysmith.ai/2026/Sep/4/hyprland/ | 2026-09-04T07:10:48.000Z | Hyprland | 
A tiling window manager that ships with [Omarchy](https://andysmith.ai/2026/Sep/4/omarchy/).
I want to try it. On the Mac I really miss this kind of approach. I've tried a bunch of them over the years (the ones that stuck ... | https://hypr.land | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/ff22dd87681da546e07c0a584b070b169e2cd4c2/hyprland.webp",
"text": "image"
}
] | [
"hyprland",
"nixos",
"tiling-window-manager",
"linux",
"omarchy"
] |
site:2026/Sep/4/installing-nixos-on-a-utm-vm-on-macos | site | https://andysmith.ai/2026/Sep/4/installing-nixos-on-a-utm-vm-on-macos/ | 2026-09-04T12:04:14.000Z | Installing NixOS on a UTM VM on macOS | Setup: a MacBook Pro running macOS Tahoe 26.6.2.
Downloaded UTM from the official site: https://mac.getutm.app
Downloaded the NixOS aarch64 image: https://channels.nixos.org/nixos-26.05/latest-nixos-minimal-aarch64-linux.iso
Installed and launched UTM. Create a New Virtual Machine.

I'm still messing around with virtualization.
VirtualBuddy is great, but it only really works with macOS guests. I couldn't get Linux going: there's no way to install from an ISO, the one Debian image is down, and I... | https://mac.getutm.app | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/82626f1f10ec0b56c4128766b77dbbfb7a6ed905/utm.webp",
"text": "SCR-20260904-rawl"
}
] | [
"virtualization",
"utm",
"nixos",
"hyprland",
"linux"
] |
site:2026/Sep/4/secretspec | site | https://andysmith.ai/2026/Sep/4/secretspec/ | 2026-09-04T15:48:41.000Z | SecretSpec | 
A tool from [cachix](https://github.com/cachix) that I'm slowly moving all my projects over to.
It lets you declare your secrets without keeping them in the repo.
So it's kind of a replacement for .env.examp... | https://secretspec.dev | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/5dae3836ede00b432f69e2276a0bd280da2d9feb/secretspec.webp",
"text": "SCR-20260904-rnrc"
}
] | [
"secretspec",
"secrets-management",
"nix",
"cachix",
"devenv"
] |
site:2026/Sep/5/the-pain-of-auto-mode-agents-in-omp | site | https://andysmith.ai/2026/Sep/5/the-pain-of-auto-mode-agents-in-omp/ | 2026-09-05T09:05:26.000Z | The pain of auto-mode agents in OMP | A few months ago, using an agent meant I read and approved every single action by hand.
Then the agents got smarter and the rejection rate dropped, so I ended up hitting Enter on autopilot, with rare exceptions.
Then I switched to auto-mode, where Claude Code used a classifier model to decide whether a given external... | null | [] | [
"ai-agents",
"claude-code",
"omp",
"agent-security",
"sandboxing"
] |
site:2026/Sep/5/architecture-review | site | https://andysmith.ai/2026/Sep/5/architecture-review/ | 2026-09-05T11:07:02.000Z | Architecture review | Agents do things, and we don't always understand what exactly.
Human teams can't keep up anymore. There's too much to understand and approve every change by hand.
So how do you guarantee quality under these conditions?
The answer is clear. Move the checking up to the higher levels of the system, and trust the LLM wi... | null | [] | [
"ai-agents",
"software-architecture",
"code-review",
"adr",
"architecture-testing"
] |
site:2026/Sep/6/an-auto-researcher-built-on-my-blog | site | https://andysmith.ai/2026/Sep/6/an-auto-researcher-built-on-my-blog/ | 2026-09-06T08:23:00.000Z | An auto-researcher built on my blog | The idea of an auto-researcher built on my notes has been kicking around in my head for a very long time.
Today I stumbled on the domain smith.wiki completely by chance, and it felt like a perfect fit for what I want. I bought it for two years (it came to a bit over 100 bucks, which I think is a steal for something th... | null | [
{
"type": "image",
"url": "https://files.andysmith.ai/img/6d7cee6e28b5e9a9d609b2fa8c385db59bc9d425/an-auto-researcher-built-on-my-blog.webp",
"text": "SCR-20260906-nows"
}
] | [
"ai-agents",
"research",
"note-taking",
"local-first",
"privacy"
] |
site:2026/Sep/6/ephemeral-agents | site | https://andysmith.ai/2026/Sep/6/ephemeral-agents/ | 2026-09-06T11:16:11.000Z | Ephemeral agents | My thinking about the automod agent problem (https://andysmith.ai/2026/Sep/5/the-pain-of-auto-mode-agents-in-omp/) came down to a few hypotheses I want to test.
The main idea is that an agent with physical access to secrets will get at them sooner or later. So least privilege is the first thing you have to build here.... | null | [] | [
"ai-agents",
"sandboxing",
"least-privilege",
"nix",
"orchestration"
] |
site:2026/Sep/9/an-interest-profile-as-a-product | site | https://andysmith.ai/2026/Sep/9/an-interest-profile-as-a-product/ | 2026-09-09T08:32:10.000Z | An interest profile as a product | An interest profile as a product.
Social networks collect a ton of information about us. They know us better than we know ourselves.
I'd like access to that profile, so my agents could know me better.
How would you get this profile out of the social networks? What formats are there for storing it? What data structur... | null | [] | [
"ai-agents",
"social-media",
"privacy",
"data-portability",
"open-source"
] |
site:2026/Sep/9/long-lived-agents-vs-ephemeral-agents | site | https://andysmith.ai/2026/Sep/9/long-lived-agents-vs-ephemeral-agents/ | 2026-09-09T08:33:04.000Z | Long-lived agents vs ephemeral agents | Ephemeral agents (https://andysmith.ai/2026/Sep/6/ephemeral-agents/) are handy for running by hand. When I want to get something done (and I don't want to give the agent full access to my machine), I can describe a sandbox, create a session, do the work (and rebuild the sandbox mid-way if I need to add something), then... | null | [] | [
"ai-agents",
"sandboxing",
"prompt-injection"
] |
site:2026/Sep/9/zeno-an-always-on-orchestrator-for-ephemeral-agents | site | https://andysmith.ai/2026/Sep/9/zeno-an-always-on-orchestrator-for-ephemeral-agents/ | 2026-09-09T08:34:51.000Z | Zeno, an always-on orchestrator for ephemeral agents | In https://andysmith.ai/2026/Sep/9/long-lived-agents-vs-ephemeral-agents/ I laid out the problems with long-lived agents.
With Zeno I tried to solve them by pulling the orchestrator out into its own layer. The orchestrator handles talking to the communication layer and the processing logic. The agents themselves can s... | null | [] | [
"ai-agents",
"orchestration",
"lisp",
"mcp",
"zeno"
] |
site:2026/Sep/10/trying-zulip-instead-of-discourse | site | https://andysmith.ai/2026/Sep/10/trying-zulip-instead-of-discourse/ | 2026-09-10T16:13:49.000Z | Trying Zulip instead of Discourse | I'm trying out Zulip as a replacement for Discourse for work notes and talking to agents.
I figure my main flow is writing notes, short and not so short, and then sending them to agents to work through. They'll comment on them, save things to the wiki, publish some as blog posts (I'll drive that part myself by posting... | null | [] | [] |
site:2026/Sep/13/test-post | site | https://andysmith.ai/2026/Sep/13/test-post/ | 2026-09-13T15:52:47.000Z | Test post | Test post. Trying out publishing posts from Zulip via Zeno. | null | [] | [] |
site:2026/Sep/14/reflection-dev-and-zeno | site | https://andysmith.ai/2026/Sep/14/reflection-dev-and-zeno/ | 2026-09-14T06:46:10.000Z | Reflection.dev and Zeno | I've spent years building and running production infrastructure at big companies with NixOS. And at night I write LISP for fun. Why not combine the two?
That's how Reflection.dev came about. It's an infrastructure framework for AI native companies, built on top of Zeno, a self-evolving agent orchestrator written in Cl... | null | [] | [] |
site:2026/Sep/14/social-media-strategy | site | https://andysmith.ai/2026/Sep/14/social-media-strategy/ | 2026-09-14T09:12:16.000Z | Social media strategy | I'm working out my strategy for publishing on social media, built around maximum openness (the open garage door principle: https://notes.andymatuschak.org/zCMhncA1iSE74MKKYQS5PBZ).
1. Private chat in Zulip. This is where I think and talk to my agents. It's my main interface, and the agents do everything else. It's the... | null | [] | [] |
End of preview. Expand in Data Studio
Andy Smith — Posts
Every public post by Andy Smith, collected in one open dataset. Why: https://andysmith.ai/2026/Nov/11/post-dataset/
The dataset is rebuilt from source on every change; each version is a commit in this repository.
Sources
| Platform | File | Status |
|---|---|---|
| andysmith.ai | data/site.parquet |
complete, updated on every publish |
Social platforms will be added later.
Fields
One row is one publication on one platform.
| Field | Type | Meaning |
|---|---|---|
id |
string | Stable ID prefixed with the platform, e.g. site:2026/Nov/11/post-dataset |
platform |
string | site |
url |
string | Public URL of the publication |
published_at |
string | Publication time, ISO 8601 UTC |
title |
string, nullable | Title, where the platform has one |
text |
string | Body as Markdown; links and images use absolute URLs |
link |
string, nullable | External URL a link post comments on |
media |
list of {type, url, text} |
Attached media; text is its text form (alt text, transcript), nullable |
tags |
list of string | Author-assigned topics |
License
CC BY 4.0. Attribute to Andy Smith with a link to https://andysmith.ai.
- Downloads last month
- 142