Private deployment, white labeling, members and usage controls, plus full auditing of Agent execution in Ultimate.
Community runs on your own servers. Standard adds enterprise identity and model governance. Ultimate adds management and auditing of everything Agents do.
| Capability | CommunityCEDeploy and run it yourself | Enterprise StandardStdAdds identity and model governance | Enterprise UltimateUltAdds Agent execution control and audit |
|---|---|---|---|
| Platform | |||
| Private deploymentRuns in your own environment, on a single server or a cluster | Self-hosted | Included | Included |
| Lead Agent & task orchestrationOne entry point takes requests, splits them into tasks and assigns them, and returns blockers to the owner | Included | Included | Included |
| Evidence-based acceptanceEach delivery comes with screenshots or a screen recording, and counts as done only after review | Included | Included | Included |
| Agent groups & external executorsSequential, parallel and debate modes, with external executors such as Codex and Claude Code | Included | Included | Included |
| Self-evolution & white-box memoryOne correction becomes a structured, viewable and editable lesson that keeps improving with use | Included | Included | Included |
| Channel integrationsFeishu and DingTalk, reusing your existing work accounts and message channels | Included | Included | Included |
| Enterprise & delivery | |||
| White labelingReplace the product name, logo and in-app branding so the product ships under your own brand | Not included | Included | Included |
| Identity, permissions & usageGrant access by department and role. Look up and export quotas, credits and token usage per user. | Not included | Included | Included |
| Model management & routingConfigure model availability, gradual rollout and pricing, with failover across upstream providers | Not included | Included | Included |
| Ultimate only | |||
| WeChat integrationWeChat, reusing your existing employee accounts and messages | Not included | Not included | Included |
| Structured Agent run tracesEvery task, action and parameter is stored as structured data, searchable and replayable by user and time | Not included | Not included | Included |
| Command-level controlsBlocklisted commands and privilege escalation are stopped, with the same rules in cloud sandboxes, local sandboxes and local machines | Not included | Not included | Included |
| Execution boundaries & network isolationLimit readable and writable paths and environment variables, keep tasks inside the sandbox, and switch network access on or off with a domain allowlist | Not included | Not included | Included |
| Sensitive content filteringFilters both input and output with built-in rule templates. Hits are logged, and you can maintain your own word lists or plug in moderation APIs from Baidu, Tencent or NetEase Yidun. | Not included | Not included | Included |
| Admin activity logEvery admin action is logged and exportable, including permission changes and sensitive operations. A separate auditor role reviews what admins do. | Not included | Not included | Included |
| Ready for your own modelAIStart collecting your private data now to post-train your own model, for better results at lower cost | Not included | Not included | Included |
Leave your details. We'll start by confirming your deployment environment, management requirements and audit scope.