tylers-blogger-blog

Local LLMs don't get the hype cycle love that cloud chatbots do. They're not always shiny, they sometimes take a little setup, and they won't magically know your company's "house style" on day one. But in real developer life-where you bounce between a repo, a ticket, a terminal, and a dozen tiny decisions-local LLMs are quietly doing the work that matters: reducing friction, keeping you in flow, and letting you ask "dumb" questions without social or security overhead. If you've mostly tried LLMs via a browser tab, you've experienced the "helpful when it's helpful" version. Local LLMs are different. They're closer to a power tool: predictable, always available, and customizable to your exact workflow. The productivity win isn't "smarter answers"-it's fewer context switches Developers don't lose hours because they can't think. They lose hours because the day gets shredded into micro-interruptions: swi...

It started like any other "quick deploy" that absolutely shouldn't have been quick. We'd shipped a small feature behind a flag, everyone was feeling smug, and I was already planning my post-merge snack. Then the alerts hit: latency climbing, error rate spiking, and a weird pattern-only one region was melting down, and only for a subset of users who'd never touched the new feature. And that's when our AI agent-internally nicknamed "Patch"-walked into the room like it owned the place. The incident: 47 minutes of chaos (and one calm agent) Our agent wasn't a "magic button." We'd wired it into three things: read-only observability (logs/metrics/traces), our runbook repo, and a tightly scoped toolbelt (query dashboards, open PRs, suggest rollbacks, and draft incident updates). No direct production writes. No cowboy stuff. At 12:18 AM, it posted in Slack: "Spike correlates with deploy `api-7f3c` but only in `eu-west`." "Mos...

I didn't "discover" AI agents in a lightning-bolt moment. It was more embarrassing than that. It happened on a Tuesday afternoon when I was wrestling with a refactor I'd been postponing for weeks: a tangle of conditional logic, half-duplicated helper functions, and comments like "TODO: make this not terrible." I asked an AI agent to draft a cleaner version expecting something... fine. Instead, it handed me code that was not only cleaner, but also easier to test, easier to read, and annoyingly consistent. The moment it out-coded me The agent's first win wasn't clever algorithms-it was basic engineering hygiene. I fed it a simplified prompt: "Refactor this function to be easier to test; keep behavior identical; add unit tests." The before-code had a long function doing validation + parsing + persistence + logging. The agent responded with: Small pure functions (validate, transform, persist) Dependency injection for the DB and logger Tests tha...

Most "productivity advice" for developers is really just a shopping list: a new IDE plugin, a new task app, a new AI tool. But the real secret is protecting your focus like it's production infrastructure. Productivity isn't speed-it's fewer context switches. Every time you jump from coding to Slack to Jira to "quickly" checking a log, you pay a hidden tax: it can take 10-20 minutes to fully regain depth. Try this: schedule two 60-90 minute deep-work blocks per day. Before you start, write a one-sentence goal ("Make the failing test deterministic") and a tiny definition of done ("Test passes 20 runs locally"). Then turn off notifications, close browser tabs, and keep a scratchpad for interruptions so they don't hijack your brain. If you want a broader framework, this guide on developer productivity tactics is a solid reference. The surprising part: the best developers don't code nonstop-they design their day to make deep work ...

Let's be real: I jumped on the AI code review bandwagon thinking it'd be a magic wand. Turns out, it was more like a noisy, slightly confused roommate who kept suggesting I wear socks with sandals. The AI would flag 'inefficient' loops that were actually optimized for readability, or miss critical context about legacy systems we were touching. One time, it told a junior dev to 'remove all comments' because it thought they were 'unnecessary noise'-yikes, that comment explained a 10-year-old workaround! We spent hours chasing down these false alarms instead of shipping features. So we did the unthinkable: we turned off the AI. We focused our human energy on what mattered : architectural risks, security gaps, and clarity for the next person reading the code. We started asking 'Does this solve the user's problem?' instead of 'Does the AI like this indentation?'. The result? Our review cycle dropped from 2 days to under 1.5 days. Why? We s...

First, check out Continue it's a plugin for VSCODE that has had me inspired to the end and back. Currently my big focus is using anthropics claude which allows me to use an LLM by a company that is eager to make it safer for humans, and I'm all about that kind of mission because we humans have a lot of big problems we face. Like Climate Change, Homelessness, and much more that I don't need to go into in this blog. With AI, a lot is possible, and a lot is at stake. Jobs for 1 are on the chopping block, however we need to understand this is a TOOL, and being sold as a big axe that can do it all on said chopping block. Now, it's time to get this 'convo with llm on web' behind us and start having the convo inside of VS CODE. Here's a screenshot of VS CODE using Continue, talking to Claude, and me, Tyler Garrett, creating  figjam alternative . I think we need to stay on top of this AI infusion, make sure we are staying up with this helpful side of AI, and sure ...