How teams run agents that need Docker, browsers, databases, or a GPU
michaelmanleyhypley · reddit · 2026-07-28
The post asks how teams run agents that need a full machine: Docker, databases, browsers, multiple repos, system packages, or even a GPU.
It frames the real problem as environment orchestration, not just model choice, and describes a setup that launches a fresh machine, runs the job, returns logs and files, caps spend, and then destroys the machine. The author is asking what others use in practice: local machines, custom VMs, or services like E2B.
More from coding & agent
- How modern AI agents plan tasks step by step before they execute — goyalshaliniuk · 2026-07-28
- How modern AI agents plan tasks with goal setup, memory, tools, and verification — goyalshaliniuk · 2026-07-28
- OpenMinis puts a runnable AI agent sandbox on the phone with deep system access — aigclink · 2026-07-28
- Moonshot’s Kimi K3 ships as a 2.8T open-weights MoE with 1M-token context — Latent Space · 2026-07-28
- PAJAMA distills LLM judges into programs and matches 13B-scale evaluators — sprocket-lab · 2026-07-28
- HumanLayer argues AI coding agents need full software-factory feedback loops — AxSaucedo · 2026-07-28