Text post by eggbrain

I'm curious how people on HN are using AI day to day, and came up with a list of questions I thought could be interesting to know. No need to answer every question of course, just wanted to get a gut feel how people are using AI.

1) What AI subscriptions are you personally paying for? (Claude, ChatGPT/Codex, Cursor, etc)

- Which tiers?

- Have you purchased multiple subscriptions from the same provider (eg two Claude accounts)?

2) Are you using models outside (or instead) of those subscriptions? For example, pay-as-you-go APIs, OpenRouter, or locally hosted models.

- Which models and providers?

3) What coding harnesses and interfaces are you using? Are you sticking with the default CLIs or IDE integrations, or using alternatives like OpenCode, Pi, Aider, Muse Code, or custom-built harnesses?

- What made you switch harnesses (if you switched)?

4) Are you using any agent orchestration or swarm tools?

- For example, Munder Difflin, Gas Town, Ruflo, or something custom.

5) Are you using broader agent platforms like OpenClaw / Hermes Agent / Grok bot / Dots / Muse?

- If you are using them, what are your main use cases?

-----

I know my workflow has differed significantly from a year ago, and so it would be interesting to capture a snapshot today of what people are using and why.

Ask HN: What's your current AI agentic setup?

An Ask HN thread asking users about their AI agent setups, including paid subscriptions, models, coding harnesses, orchestration tools and agent platforms.

What HN thinks

A census of subscription stacks, where the median developer pays three AI companies and still runs out of tokens by Thursday.

Stock Claude Code is enough
Many say a single Claude subscription with Opus and the default CLI or desktop app, letting it spawn its own subagents, beats elaborate setups. Their advice is not to overthink it.
Cheap open models via OpenRouter
A sizeable group leans on GLM, DeepSeek, Kimi and Qwen through OpenRouter, OpenCode Go or Z.ai. They use them for cost, speed, or to get around refusals from US frontier models.
DIY harnesses and orchestration
Others build custom orchestrators, kanban-driven agent workflows, or minimal harnesses like Pi with herdr to run many agents in parallel. A few use local models for privacy.

Consensus: Subscriptions are widely seen as far better value than API pricing, and Claude Code with Opus is the default for most. Swarm and agent platforms like OpenClaw and Hermes get lukewarm reviews.

Recommended in the comments

Herdr ↗ (tool): It notifies you when agents finish or need input, and several commenters abandoned their IDE setups for it.
Pi ↗ (tool): It is lightweight and works well with small context windows and local models.
OpenCode (tool): OpenCode Go is cheap and has generous limits for open models.
OpenRouter (service): It is an escape hatch for trying cheap or less restrictive models.
Hermes Agent (tool): One commenter found it far more stable than OpenClaw for long-running work.
lazygit (tool): It lets users stay in the terminal alongside their agents.