valueIQDOCS
Open valueIQ

Guides · Workspaces and admin · Security and data

Security and data

Your deals, value models and uploaded documents describe how you sell and what your customers told you. This page explains how valueIQ keeps that data inside your workspace, who can reach it, what happens to private value models, and the limits you should know about before relying on the output.

Workspace isolation

Every piece of data in valueIQ belongs to a workspace. Row-level security is enforced on every database table, so a query only ever returns rows the signed-in user's workspace is allowed to see. Identity is tied to one canonical profile per user, and storage access for documents is keyed to that same profile.

Switching workspaces switches everything: deals, products, business cases, credit balance and plan.

Roles and access

Access inside a workspace follows a four-role hierarchy: Owner, then Admin, then Member, then Viewer. Deals add their own layer, with an owner, members and read-only access. See Members and roles for how to assign them.

Access through the MCP Server and the REST API runs under the same rules. Each request is issued a short-lived token for the user making it, and every query runs in that user's row-level security context. A privileged service account is never used to read or write your data.

Authentication

You sign in with email or SSO. API endpoints are authenticated with signed tokens. The MCP Server uses OAuth 2.1, so there is no static credential to copy or store: the first time you use a valueIQ tool from an assistant, the client opens a browser tab for you to sign in and approve the connection.

REST API keys, available in beta from the Developer tab in Workspace Settings, carry scoped permissions and are rate limited.

Documents you upload

Documents uploaded into a deal or product are stored privately and served through signed URLs. Storage permissions map to your profile, so another user cannot fetch your files by guessing a path. Parsing of PDF, Excel, PowerPoint and Word files happens before the content reaches the model, and images and scanned PDFs are read with OCR.

CRM credentials

Credentials for a connected CRM are encrypted, and each connection holds its own separate key. The CRM integration currently in beta is SugarCRM.

Audit trail

Every credit transaction is recorded in a ledger, and every change to a business case (regeneration, variable edit, chart change, brand update) writes a new version with a change log that you can restore. MCP calls are logged per request. Credits spent on a workflow that fails are refunded automatically.

Public and private value models

valueIQ treats two kinds of value model differently.

A public value model is generated from publicly available data only: a company URL, a pricing page, public information. Anyone could build it from the same sources, and valueIQ treats it as public.

A private value model is what you have the moment any internal context is added: deal data, customer names, proprietary assumptions, internal benchmarks or methodology configuration. The model flips to private at that point, explicitly. Private models are your intellectual property. They are not shared with any model for training, and valueIQ does not access them. This is documented in the terms of service.

MCP and assistant privacy

When you connect Claude, Cursor, VS Code Copilot or ChatGPT to valueIQ, the assistant runs with your identity and your workspace permissions. It sees exactly what you see in the app: nothing from other accounts and nothing outside your workspace scope. Deal data passed through an MCP connection is not used for model training.

To revoke a connected client, go to Workspace Settings, open the Developer tab and remove the connection. The client loses access immediately.

Current limitations

  • Review the output. All AI-generated analysis should be checked by a person before it is used in a high-stakes pricing or value decision.
  • English works best. Analysis quality is highest for English-language content.
  • One currency per deal. Multi-currency support is planned.
  • Pricing analyses are snapshots. A pricing intelligence report reflects the page at the time it ran. There is no automated monitoring of price changes yet.
  • Built for desktop. The interface is designed for desktop and laptop use. Tablet and mobile support is partial.

Tip. If a security team needs to review the data policy before a pilot, point them at the terms of service and this page. The public and private model distinction is usually the part they want to see in writing.