Environments
Environments give AI agents a persistent identity — browser state, file storage, memory, skills, and scheduled tasks — so they build on prior sessions instead of starting over.
Persistent Browser
Cookies, history, bookmarks, and extensions carry over between sessions.
Agent Memory
Persistent instructions and context that the agent carries across every session.
Scheduled Tasks
Recurring automations that run on your schedule, with pause/resume controls.
Add skills to teach the AI how to interact with this environment
File Storage
A dedicated filesystem per environment with 5 GB of managed storage.
Configurable Skills
Add reusable skill modules that teach the agent specific workflows.
Full Desktop Access
Every environment runs on a real desktop with native OS access and machine-level interaction.
Persistent Browser
Cookies, history, bookmarks, and extensions carry over between sessions.
File Storage
A dedicated filesystem per environment with 5 GB of managed storage.
Agent Memory
Persistent instructions and context the agent carries across every session.
Configurable Skills
Add reusable skill modules that teach the agent specific workflows.
Scheduled Tasks
Recurring automations on your schedule, with pause/resume controls.
Full Desktop Access
Real desktop with native OS access and machine-level interaction.
The Gap
Frameworks handle the agent. Browsers handle the session. Nobody handles what persists between them.
Market Position
The agentic browsing ecosystem has matured rapidly. But look at where each solution sits.
Managed and ephemeral. Browserbase, Hyperbrowser, and Steel provide cloud-hosted browser sessions optimized for AI agents. Sessions are isolated, recordable, and scalable. But every session starts fresh. No persistent state. No files carry over. No memory accumulates.
Stateful and self-hosted. OpenClaw bundles a browser, file system, memory, and skills into a persistent agent environment running on your own machine. It proved the model works. But you provision the server, install Node 22+, configure the gateway, set up Chrome with X11/VNC, manage Docker, and monitor uptime.
The empty quadrant is managed and stateful. A persistent agent identity — browser state, files, memory, skills, triggers — delivered as a managed cloud service with zero infrastructure burden. That's WebRun Environments.
Beyond the Browser
Cloud browsers give your agent a browser. WebRun gives it an entire desktop. File managers, native applications, memory, multi-window workflows — everything a human uses to get real work done.
A browser is a tool. A desktop is a workspace.
Build vs. Buy
You need an agent that logs into Salesforce daily, exports leads, and sends a Slack summary. Here’s what deployment looks like.
POST /environments — create environment. Sync Chrome profile from your local browser. Salesforce session transfers automatically.pol_crm_safety automation policy. Full desktop environment enabled.Governance
Automation Policies enforce rules at the infrastructure layer — not in the prompt. The agent proposes, the policy evaluates, enforcement happens before the browser moves.
Integration
Attaching an environment to a session takes one field. No rearchitecture. No new SDK.
curl -X POST https://connect.webrun.ai/start/start-session \ -H "Authorization: Bearer YOUR_API_KEY" \ -d '{ "environmentId": "env_abc123", "policyId": "pol_xyz789", "initialTask": { "taskDetails": "Check CRM for new leads and update the pipeline", "startingPoint": "https://crm.example.com" } }'
Use Cases
Infrastructure
The agent stack has matured at every layer except one: the stateful infrastructure between sessions. Authentication, files, memory, scheduling, governance — teams rebuild this from scratch for every deployment.
Environments collapse all of it into a single managed primitive. You don't build the infrastructure. You build the flow that generate leads.
Give your agents a persistent identity. Free trial. No credit card. Two-minute setup.
Free trial · No credit card · 2-minute setup