Recoverability improves safe restarting for long ai tasks
Recoverability as a System Primitive for Long-Horizon AI Agents
Artificial Intelligence
Summary
Sometimes AI programs get interrupted during long or multi-step jobs, like editing files or using tools. Restarting these tasks often means redoing work, but continuing from saved points can cause errors if the saved state isn’t reliable. The authors explain how to make restarting decisions explicit and safe by verifying the saved state before continuing work. This ensures AI agents only resume tasks from trustworthy points, preventing hidden mistakes from creeping in.
What this means in practice
- •For ai system engineers: Build AI agents that safely resume interrupted multi-step tasks with explicit recovery controls to prevent hidden errors.
- •For devops platform teams: Integrate recoverability as a primitive in automation tools to manage state restorations with policy validation and error detection.
Authors
Zhihui Zhang, Wei Liu
Abstract
AI agents can be interrupted while editing files, calling tools, or carrying out multi-step tasks. Restarting repeats completed work, but continuing from unverified or outdated progress can carry earlier errors forward. A saved state is not necessarily a suitable place to resume. We introduce recoverability as a system primitive that makes reuse an explicit decision: select a supported starting point and a permitted recovery action, or withhold automatic continuation. Its behavioral contract binds that choice to supporting evidence, execution, and independent checks. A reference architecture connects persistence, validation, and control, with complementary runtime instances testing distinct responsibilities. Four deterministic and 20 paired file challenges demonstrate that accurate restoration and successful completion can conceal disallowed starting points. Progress controls attribute retained work to shared restoration. Event-time tests show that permission must also constrain the action, and that independently held policy evidence can expose violations even after an effect occurs. These findings establish why recovery decisions need their own evaluation, beyond restored bytes and final task success. Within supplied policies and a declared trust model, the contribution is a common, testable interface for retaining justified progress and making the conditions for its reuse explicit and enforceable.