Agent work can now begin without a live request
OpenAI has introduced scheduled automations for the Agents API, giving applications a native way to begin agent work on a recurring cron schedule or at a specified future time. A schedule can create a fresh session for independent work or deliver instructions to an existing session when the task depends on earlier context. The schedule is stored separately from the agent’s immediate turn.
The distinction matters for long-running products. A new session limits accumulated context and can simplify isolation, while an existing session preserves preferences and earlier findings. OpenAI also recommends delegating repeated work to subagents when appropriate so the main session retains the goal and concise outcomes rather than every detailed trace from every scheduled run.
Targets and invocation records make state explicit
Each scheduled occurrence creates an invocation record with its intended time, source, status and eventual session identifier. Delivery can wait if an existing session is busy, and a completed invocation means the target accepted the work rather than proving the agent finished it. Applications therefore need to follow session events and present the actual outcome separately from schedule delivery.
Periodic schedules use five-field cron expressions with an IANA time zone, while one-time schedules use a future Unix timestamp. Developers can create, update, pause, resume, manually invoke or delete schedules. Agents can also manage follow-ups in their current session when the automations tool is enabled, though the documentation limits agent tools to schedules created within that session.
Scheduling increases the need for operational controls
Native scheduling removes the need for some external cron and queue plumbing, but it does not remove product responsibility. Teams still need idempotency, deadline handling, retry policy, notification controls and a way to prevent obsolete work from running. A recurring research or monitoring task should compare with prior state and avoid repeating actions merely because a timer fired.
The strongest use cases are checks and follow-ups where context, tools and timing can be defined clearly. Applications should show the user what is scheduled, when it will next run and how to stop it. Audit records should connect the schedule, invocation, session and any external action. That chain is essential when autonomous work continues after the original conversation has ended.
What teams should validate before production
For implementation, teams should record the exact model, software, dataset, connector and policy versions behind every result. Representative evaluation must include normal work, boundary cases, deliberate misuse and component failure, with outcomes separated by user role, data sensitivity and consequence. A strong average should never hide a serious failure in a critical path. Changes should pass the same regression set before promotion, and operators need a documented fallback that preserves evidence rather than silently broadening access or lowering quality.
Procurement and planning should account for the human work required to review exceptions, maintain controls and investigate incidents. Published performance or efficiency claims are useful starting points, but production confidence comes from evidence collected in the organisation’s own environment. A named owner, measurable acceptance threshold, cost model and tested shutdown path turn a promising release into a service that can be governed over time. Where decisions affect safety, rights, money or regulated records, accountable professional review remains essential.
Security and operational evidence
Security review should map every identity, credential, network destination, stored artefact and administrative access path involved in the workflow. Logs need enough context to reconstruct who initiated an action, which evidence was consulted, what the system proposed and what ultimately executed. Sensitive prompts and traces should not be retained merely because they are useful for debugging. Retention, redaction and access must follow the underlying data classification, with a practical process for correcting or removing material when policy or law requires it.
Operational monitoring should measure quality and service behaviour together. Latency, throughput and availability matter, but so do unsupported claims, missed constraints, unsafe tool calls and the rate at which people overturn recommendations. Teams should rehearse dependency outages and partial failures before unattended operation. If a model, connector or external service becomes unavailable, the workflow should fail predictably, preserve its state and tell the user what remains unfinished. Clear degradation is safer than an agent improvising around a missing control.
Transparency should scale with adoption
Organisations should explain the scope and limitations of the deployed capability to the people affected by it. Users need to know when automation is involved, how to challenge an outcome and where a human can intervene. That transparency improves feedback and prevents a pilot result from being interpreted as blanket approval for broader autonomy. Expansion should follow demonstrated evidence, not novelty or pressure to match a competitor announcement. The durable advantage comes from dependable integration, careful measurement and learning that survives the next model or product update.
Finally, teams should schedule a review after real usage rather than treating launch as the end of governance. The review should compare expected benefits with observed quality, cost, incidents and user behaviour, then retire workflows that do not justify their complexity. A service that works well can expand through the same evidence-based process. A service that fails should be narrowed or stopped without losing the records needed to understand why. This feedback loop is what turns experimentation into accountable operational capability.