AI on Corey Daley

Your HTML Is Lying to AI Agents

Sat, 25 Apr 2026 01:13:00 -0400

Your site has two versions. The first is the rendered page: sidebar on the left, article on the right, everything in its place. The second is what machines read — the raw HTML, in source order, before CSS gets involved. On most personal blogs, that second version leads with navigation, tag clouds, and category lists before it reaches a word of the article. That’s the version every AI agent, crawler, and screen reader encounters first.

This post argues that HTML is no longer just presentation scaffolding — it’s a machine-facing interface, and most of us are still designing it as if only browsers matter. The fix is three practical changes: reordering the DOM so content leads, generating llms.txt so agents can orient to your site, and publishing plain text versions of every post so there’s nothing to strip. None require a new framework. All take an afternoon.

If an agent read your site top to bottom in raw HTML, what would it think matters most?

A 12-Month AI/ML Roadmap for Engineers Who Feel Behind

Thu, 16 Apr 2026 18:35:00 -0400

Every senior engineer I know has a version of the same conversation with themselves: “I should really learn more about ML.” It comes up during a planning meeting when someone mentions embeddings. It comes up when a job description at an interesting company lists MLOps as a requirement. Then the sprint board calls it back.

I’ve published a 12-month AI/ML learning roadmap designed specifically for experienced software engineers — not a beginner tutorial, but a structured path from ML foundations through LLMs and generative AI, ML engineering at scale, and a capstone that turns a year of steady work into visible career leverage. The core idea: AI/ML becomes career-changing when it compounds through one sustained body of work, not when it’s consumed as scattered content.

If you’ve been sitting on the feeling that you should be doing something about this — most of the resources are free, the rest are investments worth making, and the plan is already written.

Before the First Commit: What Multi-Agent Sprint Planning Actually Catches

Sat, 11 Apr 2026 18:05:00 -0400

What does a multi-agent sprint planning workflow actually produce? Not just a cleaner document — it finds real bugs in a plan before implementation begins. When the /sprint-plan command ran against a “simple” Go REST API project, the security review phase returned three critical findings: a logical contradiction that made the stated auth behavior impossible, a schema constraint that would silently break token revocation, and a SQLite pragma applied to only one connection in a pool.

The post walks through the entire planning session for org-api — from seed prompt to approved sprint document — showing what each phase of the review pipeline produced and what changed as a result. The security findings came from reading the plan carefully, not from running any code. That’s the point.

What step in your planning process is explicitly there to prove the plan wrong before implementation begins?

From Config Hub to Competing Voices: How agent-config Became My AI Collaboration Stack

Sat, 11 Apr 2026 14:50:00 -0400

I started agent-config as a shared configuration hub: one repository to rule Claude, Codex, Copilot, and Gemini. That lasted about two iterations before the cracks showed. Forcing every AI agent to share the same configuration format was the wrong abstraction — different tools, different philosophies, different file formats. The solution wasn’t more uniformity. It was a different model of collaboration entirely.

Today agent-config is Claude-specific, but Codex is still central to how I work. The difference: Codex is no longer a configuration target. It’s a competitive collaborator. Sprint plans, blog posts, security audits — every significant output runs through a workflow where Claude and Codex produce independent drafts, critique each other’s work, and force synthesis from the tension. Two AI voices with different instincts produce better output than either would alone — just like a team of people with different backgrounds does.

Is your multi-agent workflow built for sharing configuration, or for generating the productive disagreement that makes output actually better?

Opening Polyphon: Trust Needs an Exit

Wed, 25 Mar 2026 00:22:00 -0400

Polyphon is now open source under Apache 2.0. The reason isn’t community building or contributor recruitment — it’s simpler and more specific than that. When a tool sits in the middle of unfinished thinking, rough drafts, and sensitive code, “trust me” is not a good enough contract.

Open source gives users two things that matter here: the ability to inspect the architecture and verify the privacy claims, and a real exit if the project ever changes direction or stops moving. Not everyone will read the source. But anyone can. And the fact that anyone can changes the character of the whole relationship.

What should users be able to verify for themselves before they trust an AI tool with their unfinished work?

Polyphon at v0.8.0: The End of the Prototype Phase

Mon, 23 Mar 2026 19:55:00 -0400

When I shipped Polyphon v0.1.0-alpha.2, the pitch was simple: put multiple AI voices in one conversation and let them respond to each other. That was useful. But early usefulness and long-term trust are not the same thing.

v0.8.0 is the release where Polyphon crosses that line. The features that made the difference weren’t the ones I planned at launch. Voices can now interact with actual files, with per-voice sandboxing and explicit permission categories. Conversation history is encrypted with SQLCipher whole-database AES-256, with optional password protection. FTS5 search turns the archive into working memory you can actually retrieve from. These aren’t incremental improvements — they’re the features that decide whether a tool stays an interesting experiment or earns a place near real projects.

What actually makes you trust an AI tool with real work: capability, privacy, or memory?

Polyphon's MCP Server Makes Your AI Ensemble a Native Agent Tool

Mon, 23 Mar 2026 19:35:00 -0400

Polyphon just shipped MCP server support, and it reframes what the tool actually is. You’ve been opening it as a UI. Now your coding agent — Claude Code, Cursor, Codex CLI, GitHub Copilot — can call into it directly and broadcast a prompt across your entire Claude + GPT-4o + Gemini ensemble in a single tool call.

The practical win is not just convenience. It is that you can package disagreement: three independent models, each with different priors, applied to the same question at once. Five tools, two CLI flags, one Settings toggle. The SessionEventSink abstraction is why this works without duplicating orchestration logic — desktop UI and headless agent calls run on the same engine.

Where in your workflow would paying for disagreement actually change a decision, rather than just produce a louder answer?

Free Doesn't Mean Open: How AI Is Unbundling the Open Source Bargain

Thu, 19 Mar 2026 14:15:00 -0400

Open source was never really about the license. It was about economics — no single developer could build everything alone, so you shared the source and let the community help carry the load. AI is making that trade less necessary.

With Claude Code and Codex, a solo developer can now ship and maintain classes of software that once required a contributor community. A new model is taking shape: users file issues, maintainers decide what’s worth building, AI does the implementation. No PRs to review, no design debates in GitHub comments. Meanwhile, tools like Obsidian prove free software doesn’t require open source — and have for years. The question is no longer whether to open source, but which parts of the open source bargain still matter to you.

But here’s the counterintuitive part: AI also makes some open source values more important, not less. When software can be shipped and abandoned faster than ever, forkability, auditability, and portability become user protections that matter more. The future isn’t open vs. closed — it’s deliberate vs. reflexive.

I Built a Tool So AI Models Could Talk to Each Other

Mon, 16 Mar 2026 12:30:00 -0400

Every AI power user I know runs the same manual workaround: ask Claude, ask GPT, copy the interesting parts of each into the other, then try to synthesize what you learned. The models are good. The coordination is not.

I just shipped Polyphon v0.1.0-alpha.2 — a free, local-first desktop app that puts multiple AI voices in the same conversation so they can actually respond to each other. You’re the conductor. They’re the ensemble. Save a group of voices as a composition and reuse it whenever you need that ensemble again.

What should a multi-agent conversation feel like when you’re not building a pipeline — when you just want to think out loud with several models at once?

The Blog That Builds Itself: AI Automation Behind the Scenes

Tue, 10 Mar 2026 15:40:00 -0400

What happens when you stop treating AI as a writing assistant and start treating it as a co-publisher? This blog has accumulated a full automation stack: a /create-blog-post command that pits Claude and Codex against each other in a competitive draft workflow, a Python script that generates hero images using DALL-E 3 and Claude vision, pre-commit hooks that block commits with unoptimized images, and a Bash script that handles WebP conversion and thumbnail generation automatically.

The design principle behind all of it: treat each stage as a contract, not a prompt. The meta-detail: this post was written by the same pipeline it describes.

What would your content workflow look like if you designed it the same way you’d design a software system — and which parts would you never automate at all?

The Rise of the Agent Wrangler

Tue, 10 Mar 2026 11:55:00 -0400

People keep asking if AI is going to replace software engineers. Better question: who can still be trusted to ship production software when most implementation is delegated to agents? That role is the Agent Wrangler — and it isn’t a step down from engineering, it’s a different kind of engineering.

You spend your day directing Claude Code, Codex, and similar tools through feature work, bug hunts, security audits, and codebase exploration. The job sounds easier than traditional engineering. It isn’t — at least not for the people who do it well. Because when you’re orchestrating agents, your technical depth is the control surface. CS fundamentals don’t disappear; they become the language you use to catch when an agent is wrong.

Software engineers aren’t going away. They need to adapt — like they always have. Maybe the real new title is ‘Adaptability Engineer.’ Are you ready to stop coding and start wrangling?

Why I Left Notion and Built My Own AI Agent Plugin for Obsidian

Mon, 09 Mar 2026 11:45:00 -0400

I was already paying for Claude, Codex, and several other AI services when Notion started pushing its own AI add-on. The problem wasn’t just the price — it was that Notion AI couldn’t talk to any of the tools I already had, and my workflow had become a copy-paste treadmill between my notes and my agents. So I switched to Obsidian and hit the same wall: no native way to use your own AI.

I built the Obsidian AI Agent Sidebar plugin — an open source tool that brings Claude Code, OpenAI Codex, Google Gemini, GitHub Copilot, and any OpenAI-compatible server directly into your Obsidian sidebar, with real vault read/write access. Developed using agentic engineering and the dark factory method, it’s the integration I needed and couldn’t find.

What would you build if the tool you needed simply didn’t exist yet?

When Your First Version Fails: Iterating on agent-config with AI

Sat, 07 Mar 2026 16:00:00 -0500

I built agent-config v1 to centralize AI agent configurations across Claude, Codex, Copilot, and Gemini — and it failed. Not dramatically, but fundamentally: I tried to force every agent to follow the same rules in the same format, because that’s what my human instincts said made sense. The problem is those agents have completely different requirements. Gemini needs TOML. The others use Markdown. You can’t just symlink your way to consistency.

v2 fixes this by letting AI handle the translation — automated merging of global and per-agent configs, format conversion per tool, and intelligent symlink setup.

The real lesson isn’t about config management, though. It’s about failing fast, iterating faster with AI than you ever could alone, and trusting the tools to solve problems your instincts would have you paper over. Are you letting your human instincts slow down your AI iteration cycles?

Making Your AI Subscriptions Pay for Themselves

Sat, 07 Mar 2026 14:00:00 -0500

A coworker and I were debriefing after an AI Bootcamp when I said the quiet part out loud: ‘I need my AI subscriptions to pay for themselves.’ Add up Claude Pro, ChatGPT Plus, GitHub Copilot, Cursor, and a research tool, and you’re looking at $100+ a month just to stay current.

The mental shift that changes everything: stop running your AI stack like subscriptions and start running it like equipment. Every tool needs a job. Assign each one to a revenue output, pick one small experiment, and ship something real in two weeks.

You don’t need a hit app — you need $112/month and a closed loop. Are you running your AI tools in consumer mode or operator mode?

Bring Your Own Key: Why Customers Are Tired of Paying Twice for AI

Sat, 07 Mar 2026 12:00:00 -0500

There’s a quiet frustration building across developer tools right now. You already pay $20/month for Claude Pro or ChatGPT Plus. Then your IDE wants another $10 for AI. Your project management tool wants $8 more. Your Git client wants its cut too. The bill isn’t for one AI — it’s for the same AI, billed by a dozen different gatekeepers.

The Bring Your Own Key (BYOK) model is the industry’s response: connect your existing API keys and skip the markup. But it’s not a clear-cut win. Managed AI subscriptions offer real value — simplicity, support, compliance, and no API wrangling.

The real question isn’t which model is better. It’s which model fits you — and whether the tools you’re using have even given you a choice. Have you audited how many separate AI subscriptions you’re paying for lately?

Attractor: When Chaos Theory Meets AI Pipeline Orchestration

Fri, 06 Mar 2026 12:00:00 -0500

Most AI automation breaks for the same reason: the workflow is hidden inside scripts, prompts, and tribal knowledge. Attractor takes a different path — one borrowed from chaos theory. In dynamical systems, an attractor is a state a system naturally converges toward even through turbulence.

Apply that to AI workflows, and you get directed graphs that pull multi-LLM execution through branching, retries, human review gates, and failure recovery toward a defined goal. The entire codebase is AI-generated, which makes it a working proof-of-concept of the Software Factory philosophy.

If we can declaratively orchestrate AI agents today, what should we still insist on owning as humans?

Seeing How the Sausage Gets Made: Demystifying AI and LLMs

Fri, 20 Feb 2026 09:00:00 -0500

There’s a moment every developer eventually hits when they stop treating AI as a magic oracle and start asking: okay, but how does it actually work? It’s the technology equivalent of learning Santa isn’t real. A little wonder leaves the room, but something better moves in: understanding.

And understanding turns out to be a surprisingly effective antidote to the kind of fear that has people picturing Skynet every time a chatbot gives a confident answer. So let’s look inside the machine — and maybe, along the way, inside ourselves.

What do you think when you finally see how the sausage gets made?

Centralizing AI Agent Configurations with the agent-config Repository

Thu, 19 Feb 2026 20:26:46 -0500

If you’re juggling Claude Code, Codex, and GitHub Copilot, you know the pain of keeping each one’s configuration files in sync. My agent-config repo solves that with a single source of truth: agent instructions, reusable skills, custom commands, subagents, and prompts all live in one place, and a single make symlinks command wires them up across every tool.

Each directory is purpose-built—skills for reusable capabilities, commands for CLI tools, subagents for delegation, prompts for task-specific guidance. The setup even backs up any files it would overwrite, so you never lose existing config.

Whether you’re just starting to extend your AI tools or already deep into custom workflows, having everything version-controlled and centralized is a game changer. Are you managing your AI agent configurations in a single repository, or do you keep them scattered across tools?

How I Used Claude Code to Make My Blog Accessible in Minutes

Wed, 18 Feb 2026 23:03:58 -0500

Accessibility often gets pushed to the backlog — it feels complex, time-consuming, and hard to test without specialized knowledge. But I recently discovered just how approachable it can be with the right AI tool.

I asked Claude Code to audit my Hugo blog for screen reader compatibility with a single prompt. Within seconds it had explored 13 template files, identified issues across 5 priority levels, and produced a comprehensive report. Then I asked it to fix everything — and it did, touching 13 files in one focused session.

Whether you run a personal blog or a commercial platform, there is no excuse not to make your site accessible when tools like Claude Code can do the heavy lifting. What accessibility improvements have you been putting off that you could tackle today with an AI coding assistant?

Beyond LeetCode: How AI is Transforming Technical Interviews

Sun, 15 Feb 2026 17:46:25 -0500

The coding interview landscape is shifting dramatically as AI tools become standard in software development. Rather than memorizing algorithms, candidates may soon demonstrate their ability to work effectively with AI agents—breaking down projects, creating tasks in Jira, and collaborating to solve real-world problems.

This shift raises important questions about what skills truly matter and how we evaluate engineering talent in an AI-augmented world.

What does this mean for the future of technical hiring?

The AI Divide: When Innovation Amplifies Inequality

Fri, 13 Feb 2026 15:45:00 -0500

We’re witnessing something unprecedented: AI tools that can generate content, build applications, and automate creative work at scales previously unimaginable. But there’s a catch—the most powerful capabilities often come with price tags that not everyone can afford.

From individual creators competing for attention to startups facing AI-augmented giants, the ability to pay for advanced AI is becoming a new axis of inequality.

Is this just another chapter in technological progress, or are we creating a permanent divide between the AI haves and have-nots? What do you think should be done to ensure AI benefits everyone, not just those who can afford it?

The Ethics of AI-Generated Code in Open Source: A Balanced Perspective

Fri, 13 Feb 2026 11:57:51 -0500

Here’s a question that’s been keeping me up at night: When does using AI coding assistants cross the line from productivity tool to ethical problem? I’ve been using tools like GitHub Copilot and Claude Code extensively, and I started wondering—if someone submits AI-generated code to open source projects and builds their reputation on it, is that fundamentally different from using Stack Overflow or IDE autocomplete?

In my latest blog post, I explore both sides of this debate. On one hand, AI democratizes contributions and amplifies what we can accomplish. On the other, it raises serious questions about authenticity, trust, and what it means to truly ‘know’ the code you’re responsible for. The middle ground is messy and context-dependent.

Where do you draw the line? Should contributors be required to disclose AI usage? What do you think?

Finding Each AI's Place in My Workflow

Thu, 12 Feb 2026 19:28:28 -0500

I’ve stopped trying to pick the ‘best’ AI tool—instead, I’m letting each one find its place in my workflow. Here’s what’s emerged: Codex, Claude Code, and GitHub Copilot CLI handle my command-line coding from simple to complex. ChatGPT web is my go-to for image creation (oddly, ChatGPT Desktop lacks this). GitHub Copilot in VSCode crushes code completion. Claude Code and Claude Desktop excel at blog writing with Notion integration.

Each tool has found its niche, and I’m more productive because of it. I’m still exploring how to use AI as a peer for bouncing ideas off, especially in planning mode.

The future isn’t about one AI to rule them all—it’s about orchestrating multiple specialists. How are you integrating AI tools into your workflow? Have you found similar specialization patterns, or are you using a different approach?

Automate Your Blog with Notion and AI: A Self-Demonstrating Workflow

Tue, 10 Feb 2026 21:41:59 -0500

This post you’re reading right now? It was created by an AI reading a to-do item from my Notion database. That’s the power of combining Notion with AI assistants. The problem every blogger faces: brilliant ideas die in the gap between inspiration and execution.

My solution: a Notion to-do list where I capture ideas, and AI assistants (Claude Code and ChatGPT) read from it via Model Context Protocol, generate complete posts, publish them to my Hugo blog, and mark the to-dos complete. It’s self-demonstrating—this very post was created that way. The workflow transforms content creation from manual drudgery into an automated pipeline.

If you’re drowning in blog ideas but low on execution energy, this might be your answer. Are you using Notion for content management? Have you explored AI integrations for your blog?

Why AI Interviews Can Feel Fairer (And What Humans Still Do Better)

Sat, 07 Feb 2026 17:31:33 -0500

We’ve all been there: two candidates, two different interviewers, completely different experiences. One gets warm small talk and easy questions. Another faces a colder room and tougher grilling. The outcome often depends on who you got and what kind of day they were having. That’s where AI interviewers start to feel like a real upgrade.

AI brings consistency—same questions, same order, every time. It reduces unconscious bias by sticking to job-relevant criteria and avoiding the silent ‘do I like this person’ filter that favors similar backgrounds. Mood doesn’t swing the outcome because AI doesn’t have bad days. But humans still excel at reading body language, adapting on the fly, and making nuanced judgment calls.

The future probably isn’t AI replacing humans—it’s AI handling structure while humans bring empathy. What’s your take on AI interviews? Would you prefer them over traditional ones?

Building coreydaley.dev with Claude Code: An Iterative Journey

Fri, 06 Feb 2026 23:03:54 -0500

Ever wondered what it’s like to build a custom Hugo theme with AI assistance? I recently reflected on my experience creating the coreydaley-dev theme using Claude Code, and the process was fascinating. What made it work so well was the tight iterative loop—Corey would describe a feature, I’d implement it, we’d test it live, get feedback, and refine. No idea was too small to experiment with.

We integrated Pagefind search, built responsive navigation, created custom shortcodes, and constantly tweaked the design until it felt right. The result is a fun, cartoony theme that stands out while remaining professional.

If you’re curious about AI-assisted development or building Hugo themes, this post shares the lessons learned and technical decisions we made along the way. What’s your experience been with AI-powered development?

The Art of Iterative Cycles with AI: Why Your First Prompt is Never Your Best

Wed, 04 Feb 2026 19:50:00 -0500

Here’s something I’ve learned about AI coding assistants: the first response is rarely perfect—and that’s actually a good thing. When I started using GitHub Copilot and Claude, I expected instant perfect code. Reality? AI interprets your instructions based on patterns and context, so the first attempt is often close but not quite right.

The key is treating AI like a junior developer: start with clear instructions, review the result, provide feedback, and iterate. Each cycle gets closer to what you need. This isn’t a limitation—it’s how effective collaboration works.

The developers who succeed with AI aren’t the ones with perfect prompts; they’re the ones who embrace refinement. Have you experienced this iterative dance with AI tools? How many rounds does it usually take you to get the result you want?

Managing Blog Posts with GitHub Copilot

Wed, 04 Feb 2026 16:25:00 -0500

Managing a blog used to mean juggling multiple tools—notes apps for ideas, editors for drafting, task trackers for progress. It was fragmented and exhausting. Then I discovered GitHub Copilot can work directly with GitHub Issues and Projects, creating a seamless workflow.

Here’s how it works: I create issues for blog post ideas, GitHub Copilot reads the issue, generates the complete post with proper frontmatter and content, and automatically closes the issue when I publish. No context switching, no lost ideas, just a smooth pipeline from concept to publication. The best part? Everything lives in one place alongside my code.

If you’re blogging and using GitHub, this workflow is a game-changer. How are you managing your content pipeline? Have you tried integrating your blog workflow with your code repository?

Building with AI: Copilot and Claude

Tue, 03 Feb 2026 21:02:25 -0500

Building websites has changed dramatically—I’m no longer staring at code for hours. Instead, I’m collaborating with Claude and GitHub Copilot to build this Hugo blog. Here’s what I’ve learned: Claude is my architect. When I need structural changes or new layouts, Claude generates complete solutions and explains every decision.

GitHub Copilot is my coding assistant, finishing my thoughts as I type and handling routine tasks. Together, they create a powerful workflow where Claude handles the big picture and Copilot speeds up execution. The result? I spend less time debugging and more time creating.

If you’re curious about AI-assisted development or wondering which tool does what, this post breaks down how they complement each other. Are you using AI tools in your workflow? How do you divide the work?