In
Enterprise Agent Canvas and Governance PlatformEnterprise-level Agent workflow and governance platform · V06
Connecting to the live product service
Enterprise-level Agent workflow and governance platform

Turn business ideas into auditable agent workflows in a controlled canvas

Business personnel can transform business ideas into structured Agent workflows, natural language tasks, or background task drafts, and the production and research team can conduct governance reviews around connections, permissions, model restrictions, and release risks.

  1. 01
    Business idea registrationBusiness people fill in the workspace goals, data sources, role scope, and model policies.
  2. 02
    Canvas blueprint designGenerate and save Agent canvas nodes, wires, model calls, and connector references.
  3. 03
    Natural language task planningParse business personnel's vernacular requests into confirmable task plans.
  4. 04
    Governance reviewIndustry researchers and managers view permissions, connectors, model restrictions, background operations and approval risks.
  5. 05
    Preparing for controlled operationThe approval results, version status and rollback reports are only marked within this site and do not mean that the external system is online.
Number of canvas blueprintsNo canvas blueprint yetCount the number of agent_canvas_blueprint records in the current workspace.
Tasks to be confirmedNo tasks to be confirmed yetCount the number of records whose execution_state is "waiting for manual confirmation" in natural_language_task_plan.
high stakes reviewNo high-risk review yetCount the number of records with high risk_level in governance_review_report.
Registered data source typeUnregistered data sourceCounting the number of selected items in projectSchema.data_sources does not mean that the connection has been successful.
Latest blueprint verification statusNo version status yetRead the latest agent_canvas_blueprint.validation_status.
BUSINESS INPUT

New project

Each project represents an Agent workspace within the enterprise and is used to register business goals, data connections, model governance, permission scope, and release strategies.

Loading product-specific business fields.

Not saved
TRACEABLE OUTPUT

Workspace overview

Results, sources, failures, approvals, and rollbacks are saved by project.

0 live runs
∅

No business results yet

Please fill in the business goals, data sources, role scope and model strategy first.

PRIVACY

Data use boundary

The workspace may contain business objectives, industry and research contacts, scope of authority, data source types, and approval instructions. Do not paste customer privacy, keys, production credentials, sensitive personal information, or unmasked business data into business objectives, natural language instructions, or approval instructions. Current 10 Role Review Supplement: Expand Sensitive Data Scope: Business objectives, natural language source text, parsing intent, required data, team names, run plans, connector references, approval instructions, governance reports, audit events, rollback snapshots, and export reports may all contain personal information, trade secrets, or internal system information. ; It is prohibited to save API keys, OAuth tokens, database passwords, cookies, private keys, production credentials, or complete unmasked data samples in any project fields, natural language instructions, approval instructions, connector references, logs, model prompts, or export reports. ;Exporting Markdown reports containing sensitive fields must be restricted by role permissions, and provide desensitization or explicit confirmation of sensitive fields. ; Retention and deletion policies must be described separately by project fields, original messages, run logs, AI API requests/responses, export reports, approval records, audit events, and rollback snapshots; unknown third-party AI retention policies must not be fabricated.

RETENTION

Data retention

Workspace fields, blueprint versions, task plans, governance reports, and audit status are retained for 180 days by default; administrators can delete site records in advance. Deletion of this site's records will not affect any external systems not connected to this site.

SAFETY

Human responsibility and rollback

This product generates Agent canvas blueprints, natural language task plans, and governance review reports. It does not automatically prove that model answers are correct, nor does it connect to or modify production systems outside the enterprise without configuration, authorization, or approval. Actions involving production data, customer contact, permission changes, finance, personnel, legal, medical, safety or production operation and maintenance must be reviewed by authorized personnel.

Rollback only refers to withdrawing the canvas blueprint version, natural language task plan, approval status, governance review report and workflow stage mark in this site; if the enterprise separately connects the plan to an external system, the rollback of the external system must be executed separately in the corresponding system.