Agent concurrency is often a waiting problem
AWS and UC Berkeley have released Infinity, an Apache 2.0 research runtime designed for large populations of agents that spend much of their lifetime waiting. Conventional harnesses commonly hold a process and its memory while a tool runs, an approval arrives or an external event occurs. Infinity instead saves the conversation state, suspends execution and resumes the agent when work can continue.
The model borrows from asynchronous programming runtimes. An agent call is treated as durable work rather than a continuously occupied operating-system process. That can matter when thousands of coding, monitoring or triage agents may wait minutes or hours between active steps. Reducing idle resources could make background agent services more economical and less dependent on oversized worker fleets.
Suspension turns conversation state into durable work
Infinity is part of the Hydro project and is presented as research rather than a managed AWS service. Its significance is architectural: agent systems increasingly need queues, checkpoints, event delivery and replay semantics that ordinary request-response servers do not provide. Suspending at tool boundaries also offers a natural place to persist state and release compute while external dependencies finish.
Durability creates new responsibilities. The runtime must decide which state is authoritative, how duplicated events are handled and whether a resumed agent can safely repeat an operation after a crash. Tool calls that charge money or change data need idempotency keys and explicit commit records. Saving a conversation is not enough if external side effects cannot be reconciled.
Research software needs careful production boundaries
Long waits also make configuration drift more likely. A model, prompt, credential or policy may change between suspension and resumption. Systems should record the versions that began the task and define whether resumed work uses the old environment, migrates to the new one or stops for review. Expired authorisation must not be silently replaced with broader current access.
Infinity points towards agent platforms that behave more like workflow engines than chat servers. Organisations exploring it should start with reversible, observable work and measure recovery from interruption as carefully as throughput. Extreme concurrency is useful only when each resumed agent remains accountable, bounded and able to explain which state and permissions shaped its next action.
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.