Strong Isolation
Standard Runtime
Create short-lived, strongly isolated environments for agent tasks, pull-request previews, and untrusted execution.
Tennant
Authenticating...
Authenticating...
Isolate your agents. Deploy directly from GitHub. Run and inspect workload output with SDK-first APIs and a CLI operators already trust.
Module
State
Details
Visual
Strong Isolation
Standard Runtime
Create short-lived, strongly isolated environments for agent tasks, pull-request previews, and untrusted execution.
Persistent Isolation
State Retained
Keep filesystem and process state between sessions without dropping isolation guarantees for long-running workflows.
GitHub Deploy
Direct Import
Deploy a repository directly into a sandbox from branch or commit, then execute commands and inspect runtime output.
SDK + CLI Control
Unified API
Use TypeScript, Python, and Go SDKs or the Tennant CLI with one sandbox lifecycle model across teams.
(02.) Pricing Layer
Tennant Sandbox uses a simple model: start on Hobby, then scale with runtime usage and workspace concurrency.
Lobby
$0 / month + usage
Pro
Built for production teams
Enterprise
Security and scale guarantees
(03.) AI-First Infrastructure
Fast, scalable, stateful runtime primitives for production AI workflows.
Scale evals across parallel sandboxes with reproducible state and deterministic runtime behavior.
Run untrusted code in isolated environments with real-time output streaming and lifecycle controls.
Execute multi-step agent workflows with persistent context and API-driven sandbox orchestration.
Process large datasets in isolated compute sessions without leaking state across workspaces.
Render charts and visual outputs directly inside sandbox runtimes for analyst and agent tooling.
Support long-horizon training loops in controlled, stateful runtime environments.
Enable browser and terminal task automation in isolated sessions with controlled permissions.
(04.) Deployment Phase
Always On
Runtime Ready
Provision a runtime, import your GitHub repository, and issue SDK keys for your agents and applications in one control plane.
Tennant SystemsJournal
The category is moving fast. We are writing about persistent runtimes, GitHub-to-sandbox deploy flows, start latency, and what serious agent operators actually need.
Agent sessions break down when every run starts from zero. Persistent sandboxes keep filesystem state, toolchains, and runtime context intact so teams can resume work instead of rebuilding it.
The best deploy flow for agents looks familiar: connect a repo, choose a branch, land in an isolated runtime, and start inspecting output immediately.
Cold starts are usually a capacity planning problem disguised as a runtime problem. Fast sandboxes come from warm inventory, resumable state, and a control plane that knows what to keep ready.