Why does my agent restart from scratch when an infrastructure failure interrupts it midway?
Opportunity
Agents working on long, multi-step tasks have no first-class primitive for preserving intermediate progress when execution is interrupted. A timeout, rate limit hit, network failure, or container eviction at step 40 of a 50-step workflow restarts the entire job from scratch and re-executes every side effect that may have already succeeded. For a model that is 90 percent reliable per step, the probability of completing a 50-step task without any interruption is under 1 percent, and research shows success rates fall from roughly 80 percent at 10 steps to under 20 percent at 50 steps for infrastructure reasons rather than model capability ones. Building durable, checkpoint-based execution currently requires each development team to implement it from scratch, with no standard primitive or protocol comparable to a database transaction log.
Why it matters
Until durable execution is a first-class infrastructure primitive for agents, every increase in task length amplifies failure cost exponentially rather than linearly.
How I score the opportunity
The Opportunity Score is my own read, not a measurement: how much it hurts, how often it bites, and how little exists to solve it today. Higher means I think it is more worth building.
How much pain it causes when it shows up.
How often people actually run into it.
How little good tooling exists for it today.
More problems worth solving
Why does every AI app forget me the moment I close the tab?
AIWhy is learning a new field still gated by knowing what to ask?
AIWhy can a non-expert not verify what an AI just told them?
AIWhy do we test models on benchmarks but ship them on vibes?
AIWhy do AI agents have no memory of their own mistakes?
AIWhy can't I audit what a model was actually trained on?