Five-Level Autonomy Model for AI Coding Agents
DisturbedGoW · reddit · 2026-07-19
The author proposes a "Supervision Ladder" to distinguish five levels of autonomy for AI coding agents:
- Assist: Only autocomplete, humans handle every line of code.
- Review: Agent generates whole changes, humans review each before deployment.
- Checkpoint: Agent generates and tests, automatic gating + human interception of failures. The author believes this is the sweet spot for most real product development.
- Orchestrate: Multiple agents work in parallel, coordinated and reviewed by a senior human.
- Autonomous: Agent can run unattended, suitable for one-off exploration, not for production-critical tasks.
The author emphasizes: Demos like "run all night and wake up to a thousand commits" are cool but are essentially level 5 autonomy; once real customers or money are involved, any unsupervised error going straight to production is dangerous. The real key is control at levels 3/4: not just prompting the agent not to overstep, but structurally removing its ability to overstep. The author gives an example: they tried having the main agent "delegate" using natural language instead of writing code itself, but under high load it still took over; the effective solution was to directly remove its file-editing permissions, so it could only delegate.
More from coding & agent
- Ben Bajarin: Frontier AI is becoming the enterprise cyber-defense stack — BenBajarin · 2026-10-26
- Building AI bots on raw APIs? Devs say models are 'dumb without the harness' — madeofAjala · 2026-10-26
- HarnessMM cuts agent tokens per task by up to 88% while raising success rates — omarsar0 · 2026-10-26
- Microsoft demos intent-classifying agents built with Copilot App and Foundry, deployable across Teams and Outlook — marlene_zw · 2026-10-26
- Looking for a wearable mic with best battery life that feeds recordings to personal agents via MCP — altryne · 2026-10-26
- The identity crisis in software engineering: multiagent systems demand a new problem set — rakyll · 2026-10-26