AgentJack is the governed runtime for private AI workflow pilots. One customer, one workspace, isolated execution, human approvals, evidence trails, usage controls, and optional BYOK. Governed AI operator runtime with isolated workspaces, approvals, and metered usage.
Run private AI workflow pilots in isolated environments with approvals, evidence, usage caps, and vertical apps that map to real operator jobs.
Each customer runs in their own isolated container. Complete separation of data, models, and compute resources.
Deploy multiple AI instances per team. Scale up or down based on workload without impacting other customers.
Enterprise-grade security out of the box. Encrypted data at rest and in transit, with fine-grained access controls.
Launch purpose-built apps for computer support, reasoning, communications, creative, learning, security, and automation. One platform, focused surfaces.
From first container to production-grade AI infrastructure. Deploy isolated AI assistants that your team can actually use.
Sign up and get a dedicated container. Pre-configured with browser, terminal, and file system tools. Zero setup required.
Add API keys for the tools you use. GitHub, Linear, Slack, Notion, custom APIs. One-time setup, works everywhere.
Create agent instances for different tasks. Run multiple instances per team. Your AI infrastructure grows with you.
AgentJack beta is invite-only. Access starts with a $1 exploratory signal or an approved invite credit, then plans combine license access, included usage, prepaid credits, BYOK, and dedicated hosting where needed. Checkout is moving through Verduona WooCommerce while Sphere stays the workspace, entitlement, usage, and provisioning source of truth.
For real people testing whether AgentJack should become their operator workspace.
For founders and operators who want a persistent AgentJack workspace with regular model and workflow usage.
For small teams that need shared agents, Operator, Bridge, approvals, audit, and clearer workspace boundaries.
For dedicated hosting, BYOK/provider-owned credentials, Mothership/Infra Worlds capacity, or on-premise-style deployments.
AgentJack package checkout is planned to run through the Verduona shop: WooCommerce handles cart, receipt, invoice, refunds, and customer billing UI; signed commerce events then update Sphere, which owns workspace entitlement, usage budgets, tenant state, and provisioning. Until staging verification is green, request access instead of relying on a public live checkout.
We are opening AgentJack in controlled waves: internal candidates first, public waitlist next. Tell us what you want the agent to run for you.
Invite-only beta • Dedicated workspace • Manual review before activation
AgentJack is the runtime for governed AI work. Core provides the operating boundary; each vertical gives a focused surface for a real business workflow.
Governed runtime, isolated workspace, approvals, credits, memory, and deployment paths for real AI operator work.
Reasoning and decision workspace. Deep analysis, synthesis, and structured thinking with persistent memory.
90-minute Brain decision sprint where six expert advisors debate live options and deliver a scored recommendation.
Founder-facing knowledge surface. Capture, organize, and surface institutional memory across your team.
Automation and coding-production lane for repeatable operator jobs, implementation runs, and reviewed handoffs.
Security controls and guardrails. Manage access policies, audit logs, computer access approvals, and compliance from one place.
Gated learning SaaS for courses, cohorts, progress, AI tutors, and operator adoption. Access is authorized through Sphere learner/creator entitlements created from the Verduona shop corridor.
Shared knowledge and collaboration. A communal workspace where teams build and refine ideas together.
Call center, sales, support, and campaigns. One integrated inbox for calls, chats, emails, and outreach.
Avatar and creative generation. Produce visual and written content at scale with brand consistency controls.
Agent factory for durable specialists. Build agents that run 24/7 with their own memory, tools, and workflows.
Infrastructure and world-state lane. Manage deployments, system state, and your AI infrastructure footprint.
Czech safety vertical for BOZP/PO evidence, standalone ADR workflows tied to the current biennial ADR Agreement, radiation-protection preparation, and documented qualified-review follow-up.
Outputs organize evidence and follow-up; they are not legal advice, inspection results, certification, ADR training, radiation-protection authorization, or a replacement for qualified BOZP/PO/OZO, ADR, radiation-protection, occupational-health, medical, social-service, or legal reviewers. Qualified human review is required.
Finance and accounting operating surface for structured records, review queues, and AI-assisted back-office work.
Draft real-world venue and place-operations vertical for environments that need agents, workflows, and human oversight.