--- title: "RevSure Canvas: Product Overview, AI Architecture & Security" slug: "revsure-canvas-product-overview-ai-architecture-security" description: "RevSure Canvas Customer Security & Compliance Review Guide covers Anthropic Claude, Google Cloud, data access, AI governance, privacy, and tenant isolation." tags: ["AI Security", "Data Privacy", "Tenant Isolation"] updated: 2026-11-03T10:19:34Z published: 2026-11-03T10:19:34Z canonical: "documents.revsure.ai/revsure-canvas-product-overview-ai-architecture-security" --- > ## Documentation Index > Fetch the complete documentation index at: https://documents.revsure.ai/llms.txt > Use this file to discover all available pages before exploring further. # RevSure Canvas: Product Overview, AI Architecture & Security **Customer Security & Compliance Review Guide** ## 1. Overview RevSure Canvas is an AI-powered conversational analytics workspace that enables users to explore their marketing, sales, and revenue data through natural-language questions, generate interactive dashboards, and share insights with their teams. Canvas builds on RevSure's existing data intelligence capabilities and uses the same RevSure data access layer as Reli, including Model Context Protocol (MCP) tools, tenant isolation, and backend-enforced user permissions. While Canvas shares Reli's underlying data access and security framework, it introduces a separate AI processing architecture powered by Anthropic Claude, hosted through Google Cloud Agent Platform (formerly Vertex AI). This document outlines how RevSure Canvas works, its relationship to Reli, and the additional considerations for organizations conducting an AI security and compliance review. ## 2. RevSure Canvas and Reli — What's Shared and What's Different? RevSure Canvas and Reli share the same underlying RevSure data infrastructure and access controls. However, Canvas introduces additional AI capabilities for conversational analytics, multi-step reasoning, and interactive dashboard generation. | Component | Reli | RevSure Canvas | | --- | --- | --- | | AI model | Google Gemini | Anthropic Claude | | AI infrastructure | Google Cloud | Google Cloud Agent Platform | | Data access | RevSure MCP tools | Same RevSure MCP tools | | Tenant isolation | Backend-enforced | Same backend enforcement | | User permissions | RevSure role-based access | Same RevSure role-based access | | Model training on customer data | Not used for training | Not used for training | | AI processing | Reli AI architecture | Separate Claude Agent SDK architecture | | Analytical workflows | AI-assisted insights | Multi-step conversational analysis | | Dashboard generation | Existing Reli capabilities | AI-generated interactive artifacts | | Persistent workspace | Feature-dependent | Chat history and saved artifacts | | Collaboration | Existing product permissions | Private, public, and link-sharing options | | Automated refresh | Feature-dependent | Scheduled artifact refresh and notifications | **Key distinction:** Canvas reuses the existing RevSure data access and authorization framework but adds a separate AI runtime, model provider, code-generation process, and storage capabilities. Organizations that have already approved Reli can use the existing data access and security documentation as part of their Canvas assessment. However, the additional Canvas-specific processing and functionality may require a separate security review. ## 3. How RevSure Canvas Works RevSure Canvas combines conversational AI with RevSure's authorized GTM data to generate analytical answers and interactive dashboards. **Step 1 — User authentication** A user signs in to RevSure Canvas using their organization's configured authentication mechanism. RevSure validates the user's identity, tenant membership, and assigned permissions. **Step 2 — Natural-language request** The user submits a business question, such as: "Compare pipeline generated by marketing channels this quarter versus last quarter and create a dashboard showing the results." **Step 3 — Secure data retrieval** Canvas uses RevSure MCP tools to retrieve the relevant marketing, sales, and revenue information. Each request is scoped to the authenticated tenant and the user's assigned permissions, with access controls enforced by RevSure's backend. **Step 4 — AI-powered analysis** Anthropic Claude processes the request and authorized data returned by RevSure's tools. For more complex questions, Canvas may execute multiple analytical steps to retrieve additional relevant information before preparing its response. **Step 5 — Dashboard generation** When an interactive dashboard is requested or appropriate, Canvas uses a separate code-generation step to create charts, tables, KPIs, and other visual components. Generated artifacts are rendered in a controlled, network-isolated sandbox. **Step 6 — Results and collaboration** Users can review the response, refine the analysis, save artifacts, refresh dashboards, and share approved outputs with other authorized users. ## 4. AI Model and Hosting Infrastructure RevSure Canvas uses **Anthropic Claude through Google Cloud Agent Platform**, integrated using the Claude Agent SDK. The AI processing architecture includes: - **AI model provider:** Anthropic Claude - **Model hosting and inference infrastructure:** Google Cloud Agent Platform - **Cloud environment:** RevSure-managed Google Cloud project - **AI orchestration:** Claude Agent SDK - **Data retrieval:** RevSure MCP tools - **Artifact generation:** Dedicated code-generation process - **Artifact rendering:** Network-isolated execution environment Users may select from supported Claude models, depending on their available configuration. RevSure does not host or maintain the underlying Claude model itself. Anthropic maintains the model, while Google Cloud provides the managed inference infrastructure. ### Relationship to existing Google Cloud infrastructure Canvas continues to operate through RevSure's Google Cloud environment, using the same foundational cloud platform as RevSure's existing AI capabilities. However, because Canvas uses Claude rather than Gemini, its model-specific processing and applicable third-party terms should be reviewed separately. ## 5. Data Security and Access Controls RevSure Canvas is designed to maintain the same tenant and permission boundaries as the existing RevSure platform. ### 5.1 Tenant isolation Each customer's data is isolated within its authorized tenant environment. The AI layer cannot independently select or change the tenant context. Tenant identity and authorization are validated by RevSure's backend for each request. ### 5.2 User-level permissions Canvas does not independently grant users additional access to RevSure data. Access to data through MCP tools remains subject to the authenticated user's assigned roles and permissions. ### 5.3 Data encryption RevSure uses Google Cloud infrastructure with encryption protections for data in transit and at rest. ### 5.4 Controlled data access Canvas retrieves data through RevSure's authorized MCP tools rather than providing unrestricted access to the underlying databases. The data retrieved depends on the user's question, the available tools, and the permissions associated with the user's account. ### 5.5 Data minimization Canvas is designed to retrieve and process the information needed to respond to a specific analytical request. Depending on the use case, this may include aggregated business metrics, campaign information, pipeline data, account information, and limited professional contact details. Where supported and configured, data masking or field restrictions can further limit the information included in AI processing. ## 6. Customer Data and AI Model Training **RevSure does not use customer-provided data to train Anthropic Claude models through RevSure Canvas.** Claude is accessed through Google Cloud's managed AI infrastructure. Under the applicable published commercial terms: - Google Cloud restricts the use of customer data for training or fine-tuning managed AI models without the required customer permission or instruction. - Anthropic's commercial terms for Claude on Google Cloud restrict training on Customer Content processed through the service. - RevSure uses customer data for authorized service functionality, security, operations, and support, rather than independent foundation-model training. The data processed depends on the specific analysis requested by the user. For further details, refer to the [RevSure Canvas — Claude AI Usage & Compliance](https://documents.revsure.ai/docs/revsure-lite-ai-compliance) documentation. ## 7. Additional Functionality Introduced by Canvas Compared with Reli's existing AI capabilities, RevSure Canvas introduces several additional components relevant to security and governance reviews. ### 7.1 Multi-step AI tool calling Canvas can perform a sequence of analytical operations within a conversation. The AI may invoke multiple RevSure MCP tools to retrieve and analyze information before returning an answer. The underlying data access permissions remain enforced by RevSure. ### 7.2 AI-generated interactive dashboards Canvas supports a dedicated code-generation process that transforms analytical results into interactive dashboard components. The generated artifacts are rendered within a network-isolated sandbox that cannot make outbound requests to RevSure systems or external services at runtime. ### 7.3 Conversation and artifact storage Canvas stores conversation history and saved analytical artifacts to support continued analysis and reuse. For large analytical results, Canvas may use temporary files within an isolated session environment. These temporary files are session-scoped and are not retained in shared storage beyond the session lifetime. Conversation history, saved artifacts, and operational logs are managed under applicable RevSure retention schedules and customer agreements. ### 7.4 Sharing and collaboration Canvas supports multiple artifact visibility and sharing options, including: - Private artifacts - Public artifacts within the supported product sharing model - Link-shared artifacts Organizations should review the available visibility settings and establish appropriate internal policies for sharing AI-generated analyses. ### 7.5 Scheduled refresh and notifications Users can configure recurring artifact refreshes so that dashboards are updated using the latest available RevSure data. Supported notification options include email and Slack. These features introduce additional output-delivery workflows that may be relevant to organizational security and information-sharing policies. ## 8. Privacy, Retention, and Responsible AI RevSure applies access controls, monitoring, and governance practices to its AI processing environment. Key considerations include: **Data retention:** Conversation records, saved artifacts, and operational logs are retained according to applicable retention schedules and customer agreements. Retention configurations and relevant periods can be reviewed with RevSure. **Temporary processing:** Session-scoped temporary files used for large results are isolated and retained only for the active session. **Monitoring:** RevSure monitors AI activity for unauthorized access, excessive data retrieval, prompt injection, unsafe outputs, and suspected data exposure. **AI-generated results:** Responses and dashboards may contain inaccuracies or incomplete interpretations. Users should validate outputs before relying on them for material business decisions. **Customer governance:** Organizations can establish internal approval processes covering access, permitted use cases, data handling, and artifact sharing. ## 9. Recommended Security Review Scope For organizations that have previously reviewed and approved Reli, the RevSure Canvas assessment can focus on the additional functionality and processing introduced by Canvas. | Review area | Security review consideration | | --- | --- | | Existing data access | Confirm reuse of RevSure MCP tools and backend authorization | | AI model provider | Review Anthropic Claude through Google Cloud | | AI processing | Review multi-step reasoning and model inference | | Code generation | Review isolated dashboard-generation and rendering environments | | Data storage | Review conversation history, saved artifacts, and retention | | Data privacy | Review permitted data types, masking, and handling | | Sharing | Review private, public, and link-sharing capabilities | | Notifications | Review scheduled refreshes and email/Slack delivery | | Compliance | Review applicable Google Cloud and Anthropic terms | A previously approved Reli deployment provides a foundation for evaluating Canvas's shared security controls. **Approval of Reli does not automatically constitute approval of Canvas's additional AI model, processing, storage, and collaboration capabilities.** Organizations may complete their internal security review before enabling additional Canvas use cases. ## 10. Supporting Documentation ### RevSure Canvas — Product Documentation | Document | Description | | --- | --- | | [RevSure Canvas Overview](https://documents.revsure.ai/docs/revsure-lite-overview) | Features, workflows, conversational analysis, and interactive dashboards | | [RevSure Canvas Best Practices](https://documents.revsure.ai/docs/revsure-lite-best-practices) | Guidelines for effective questions, analysis, and dashboards | | [Claude AI Usage & Compliance](https://documents.revsure.ai/docs/revsure-lite-ai-compliance) | AI architecture, data handling, security, and compliance | ### RevSure Data Access and Security | Document | Description | | --- | --- | | [MCP Security FAQs](https://documents.revsure.ai/docs/mcp-security-faq) | Authentication, tenant isolation, permissions, and security controls | | [MCP PII Policies](https://documents.revsure.ai/docs/revsure-mcp-pii-policies) | Personally identifiable information and privacy controls | | [MCP Tool Catalogue](https://documents.revsure.ai/docs/revsure-mcp-tool-catalogue) | Available data tools and supported operations | ### Third-Party Model Documentation | Provider | Documentation | | --- | --- | | Anthropic | [Claude Models on Google Cloud](https://platform.claude.com/docs/en/build-with-claude/claude-models-on-google-cloud/google-managed) | | Google Cloud | [Anthropic Claude on Agent Platform](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/claude) | | Google Cloud | [Using Partner Models](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/partner-models/use-partner-models) | | Anthropic | [Commercial Terms for Claude on Vertex AI](https://www-cdn.anthropic.com/471bd07290603ee509a5ea0d5ccf131ea5897232/anthropic-vertex-commercial-terms-march-2024.pdf) | ## 11. Summary RevSure Canvas extends RevSure's existing AI-powered revenue intelligence capabilities with conversational analytics, multi-step analysis, and interactive dashboard generation. It shares Reli's RevSure MCP data layer, backend authorization, tenant isolation, and core Google Cloud infrastructure. The primary additions are Anthropic Claude model processing, a separate AI orchestration architecture, interactive artifact generation, persistent analytical workspaces, and expanded sharing and scheduling capabilities. These differences are documented to help customers evaluate Canvas within their existing AI governance, security, privacy, and compliance processes. For additional security documentation, configuration details, or questions about deployment-specific controls, please contact the RevSure team. ## Related - [RevSure Canvas: Claude AI Usage & Compliance](/revsure-canvas-ai-compliance.md) - [MCP Security FAQs](/mcp-security-faq.md) - [RevSure Canvas: Your Conversational Analytics Solution](/revsure-canvas-overview.md)