The Polyaxon agent sandbox roadmap
Our plans for persistent sandbox workspaces, suspend/resume, activity-aware idle suspension, and ready pools on Kubernetes—and where AX and Substrate differ.
Explore practical AI agent guides covering architecture, evaluation, sandboxed execution, tracing, security, and production operations.
Essential guides, selected for this topic.
Our plans for persistent sandbox workspaces, suspend/resume, activity-aware idle suspension, and ready pools on Kubernetes—and where AX and Substrate differ.
Design a Kubernetes platform for AI agents with clear workload boundaries, durable state, scoped access, resource controls, and end-to-end evidence.
Map agent architectures to Polyaxon jobs, services, and DAGs, with concrete workload configuration and a repeatable architecture-comparison workflow.
Design isolated Kubernetes execution for AI agents with a threat model, stronger runtimes, scoped identity, controlled egress, bounded storage, and auditable cleanup.
Run parallel coding agents in separate Polyaxon workspaces, collect candidate patches, evaluate independently, and merge only reviewed changes.
Design agent execution around durable state, safe tool retries, approval waits, and recovery, with clear responsibilities for Polyaxon and your agent framework.
Operate dynamic AI agent plans with Polyaxon execution, explicit policy checks, cumulative budgets, durable request state, and independent evaluation.
The latest guides, tutorials, and product updates.

Size an agent sandbox's working directory, set ephemeral storage requests and limits, and save the files that should survive the Pod.

Pin sessions to an agent revision, roll out a candidate to new sessions, and retire the previous revision after its work drains.

Scale an SQS-backed agent worker pool from waiting and in-flight tasks, and handle a worker shutdown without losing the task.

Host an agent controller on Polyaxon to launch training and evaluation jobs, inspect results, and choose the next experiment.

Build agent recovery around committed checkpoints, versioned inputs, safe side effects, and approval state so interrupted tasks can continue reliably.

Reconcile uncertain tool effects, preserve approval and cancellation across task changes, and compare recovery policies using final-state evidence.

Build a measured research loop around Jev or a self-hosted classifier using Polyaxon jobs, sandboxes, experiment matrices, and per-case evidence.

A design note on autonomous experimentation: fixed protocols, selection bias, independent evaluation, and durable research campaigns in Polyaxon.

Give an open model a bounded read, edit, and check loop in a Polyaxon workspace, then retain the patch and evidence behind its completion claim.

Our plans for persistent sandbox workspaces, suspend/resume, activity-aware idle suspension, and ready pools on Kubernetes—and where AX and Substrate differ.

Separate identity, network access, data, queues, budgets, and evidence when multiple teams or customers run AI agents on shared Kubernetes infrastructure.

Design isolated Kubernetes execution for AI agents with a threat model, stronger runtimes, scoped identity, controlled egress, bounded storage, and auditable cleanup.

Choose the Kubernetes execution shape for an AI agent by separating logical task identity from Pods, Jobs, Services, workflows, and sandbox lifecycles.

Run parallel coding agents in separate Polyaxon workspaces, collect candidate patches, evaluate independently, and merge only reviewed changes.

Design a Kubernetes platform for AI agents with clear workload boundaries, durable state, scoped access, resource controls, and end-to-end evidence.

Connect agent traces to checkpoint state, versioned artifacts, action receipts, and evaluations so failures lead to controlled recovery and reproducible improvements.

Define agent outcomes, prohibited actions, events, evaluators, budgets, segments, and feedback loops before implementation begins.

Operate agentic AI with explicit service ownership, platform contracts, task lifecycle rules, change control, incident response, SLOs, and financial accountability.

Contain AI agent identity, tools, memory, network access, and delegated actions with enforceable production boundaries.

Connect AI agent execution to Polyaxon projects, components, scheduling, tracking, and artifacts through a stable application-owned execution contract.

Package application-owned agent evaluators as versioned Polyaxon components with typed contracts, reusable execution profiles, and consistent result evidence.

Choose jobs, services, queues, and capacity controls for agent workloads using task duration, dependency limits, resume latency, and execution cost.

Build an agent release checklist around Polyaxon qualification jobs, versioned components, artifact reports, termination settings, and manual approval.

Build an operational foundation for autonomous agents using Polyaxon workload profiles, durable state, bounded authority, evaluation, and recovery ownership.