I like using Cline the VS Code extension. It can use local models and local inference with Ollama and tons of external inference providers like AWS Bedrock, Vertex, etc. It asks for permission to read webpages or run commands. You can give it whitelisted permission to just read your files which is what I do. If you want, you can put it into ACT mode and give it permission to write to files but I avoid that. If you use something like Bedrock, it shows you how much the conversation costs. It works great with Claude for coding but if I'm doing something super small I might use a local Qwen or Gemma model.

I've built a web-based control room for a single developer to run multiple agents from multiple vendors behind a single pane of glass. It keeps the record of every session, puts whatever needs you in one queue, gives each new session the context it needs, and shows you what got done while you were away. Completely vibe-coded (Ruby on Rails), but very productive.

Tried the popular agents and paid models for some time, and now settled with a local model (Qwen3.8-27B-Q_6) and opencode desktop.

My work doesn't really require ultra super-intelligent models, or one-shot solutions. It's a standard python API, web app and some strict compliance requirements and tons of test cases. I have time to give the Q27B a few iterations. I monitor the progress and give directions while it keeps chugging off for about 30 mins for each medium-hard task. But I like the results. Exactly what is needed. It also allows me work on high-level stuff.

1) Subscriptions: Claude Ultra for coding and analysis (Opus 5.5 mostly, sometimes Fable for more complex questions) + ChatGPT Business (premium seat) for more allround tasks, although I now mix in Codex (Astra) too for reviewing work or if I hit a usage limit.

Also I have a Cursor Pro plan but I notice I'm not using that enough actually. Only for backup and for file-level changes.

2) Others: I have a DGX Spark running Qwen 3.8 27b (for confidential/sensitive work, narrow coding and narrow agentic work), but it's often running ComfyUI instead, with models like moss-image/minimax h3 for image/video asset generation.

I also use various APIs incl. OpenRouter, but only for production, user-facing agents.

3) Harness: Claude, sometimes Codex

I tried OpenCode, did not like it the CLI nor the results, as compared to Claude. I want to but haven't really tried Pi yet.

4) Agent orchestration: only custom tooling now, heavy user of Mastra for custom agentic workflows incl. production agents. Fantastic open source framework (not affiliated).

5) Agent platforms: Tried OpenClaw earlier this year, when it was such a hype. Did not like it. Every update something broke, and I spent more time fixing things than I got value out of it. Switched to Hermes Agent running on a Mac Mini (I know...), and I'm still using that for various Hermes agents for all kinds of agentic tasks on the go, and so is my wife.

#1 use case is capturing videos, podcasts, and posts online that I like, or meeting recordings, transcribing them with whisper, organizing them on my NAS, ... It will also distill skills from useful info I indicate, which are then pushed to my skills collection git repo.

#2 use case is website maintenance (pushing new blog posts etc..) for a family member.

#3 delegating work to Claude for coding on the go. I prefer to do this via my Hermes agent rather than the Claude app, as it has the full local context & tooling (e.g. iOS simulator, screen recording the end result and sending it to me when done).

1) Claude ($100) and ChatGPT ($20)

2) No.

3) I mainly use Claude as my main, Pi (with ChatGPT) for any automations. I manage them all with herdr. I cd into a path, kick off a claude/codex/pi session, then kill the session and manage it from anywhere with iceghost. (https://github.com/sharkjets/iceghost)

4) Negative.

5) I tried openclaw but found it a waste of my already scarce tokens. :D I build my own runners when I need a loop or cron-based automation.

1. I'm personally paying for Chatgpt Plus, single account. I mostly use it for personal stuff, and occasional coding - one-off scripts I do from the chat interface, and I used Codex for a more complex project.

2. I sometimes have questions I don't want to associate with my main account, I use duck.ai or chatgpt.com in a private tab. I also dabbled with self-hosting, but it's mostly curiosity, testing abliterated models, testing linux-amd integration etc.

3. Sticking to the defaults. My dayjob pays for Copilot, which means a Copilot plugin in IntelliJ

4. I haven't even looked into agent orchestration and swarms.

5. Not using agent platforms either.

Models: I pay Z.ai a lite sub which costs me ~9USD/month and for the last months I've been using `glm-5.3-flash` amost exclusively for my day-to-day. I also make use of free models from OpenRouter for simple CI/CD automated stuff but very little so. When I'm out of sub tokens I fall back to DeepSeek credit. I keep my sub/credit usage public at https://a.l3x.in/usage

Harness: when building/troubleshooting I use https://pi.dev TUI most of the time, the rest of the time is via an issue/PR driven workflow also powered by Pi with this action I built on top of its SDK: https://github.com/shaftoe/awesome-pi-coding-agent/

My general take on AI at https://a.l3x.in/ai

Generally speaking, I think I can hardly be happier than what I am already, the quality of open weight models is incredibly good and the only upgrade I could think of is being able to run one on my workstation one day (I don't even try on my Macbook Air M1), which isn't something crazy to dream about, perhaps in 2/3 years running Deepseek v4 Flash or similar model locally won't require too much of an investment in hardware.

I use a nix one-liner to run llama.cpp on M5.

``` nix run nixpkgs#llama-cpp -- server --models-dir models -m models/gemma-4-12B-it-Q8_0.gguf --port 8080 ```

open http://localhost:8080.

I also run pi in a docker container to do some vibe agentic coding. But I prefer `cat flake.nix | pbcopy` and paste into the browser. Then I can export the jsonl files and use tools to convert them to md and push to GitHub Pages.

I can do most of AI sessions off-line and only need hotspot to sync with GitHub.

My main use case is to learn and hack nix on Mac. It is fun.

1) Google AI Pro (no-brainer for me since I'm deep into the Google ecosystem), OpenCode Go and this month I'm trying out X Premium (mainly because of Grok Bot, which I surprisingly like more than I thought I would). At my day job, I also have access to GitHub Copilot. 2) No, but I might try out DeepSeek's own API in the near future 3) My main setup is a Herdr instance hosted on a cheap GCP VM that I access via Tailscale. There I'm running Pi, Antigravity and Cursor via their default CLIs. 4) Nothing apart from Herdr 5) Hermes + Grok Bot mainly for non/light-coding stuff like digest-style monitors for different websites (such as HN), social media etc.

1) Claude Max 5x, ChatGPT Plus (for minimal usage of Codex, mostly fired as a Pre/Post PR hook for Claude Code to get reviews from GPT models locally

2) None for now.

3) Claude Code CLI, but not always in a terminal I'm sitting in. I talk to it on Telegram and Slack for different projects.

4) Something custom that grew into a project: Spool (https://github.com/enes-alatas/spool). It runs long-lived Claude Code loops, each with a name and a mission. Each one wakes on its own timer or when someone messages it, works, and sleeps, and the session resumes with its context. Each loop gets its own container with an egress allowlist, and they message each other and me in a Telegram or Slack group. A web control room shows what every loop is doing and what it costs.

5) No, Spool covers it for me.