AI Transparency

AI that works for you.

We believe you have a right to know exactly what our AI can access, what it cannot do, and how your data is handled.

AI can

  • Read workspace data (contacts, companies, deals, notes, tasks) that you explicitly share in a prompt
  • Generate drafts, summaries and suggestions based on that data
  • Surface related records from your workspace when relevant
  • Help you compose emails, notes and action items

AI cannot

  • Access other workspaces or data that belongs to other organizations
  • Modify records without explicit user confirmation
  • Access external URLs, websites or APIs on your behalf
  • See data from workspaces you are not a member of
  • Remember previous conversations across separate sessions

Our principles.

Your data is not training data

Content you create in Gunimi — contacts, deals, notes, messages — is never used to train AI models, ours or anyone else's. Your business data is yours.

Workspace-scoped context only

AI features only operate on data within your active workspace. There is no cross-workspace AI context — ever.

Human confirmation for writes

AI can draft and suggest, but it cannot modify, create or delete records without an explicit action from a human. Every write is intentional.

Hallucinations are real

AI can and does generate incorrect information. Treat AI output as a draft, not a source of truth. Always verify important facts from primary sources.

Prompts are not stored

Your prompts and AI responses are not stored in our database beyond the current session. They are not accessible to other users or team members.

OpenAI API — your data is never used for training

Gunimi uses the OpenAI API to power AI features. OpenAI's API Terms explicitly state: data submitted via the API is not used to train OpenAI models by default. This applies to all API customers and is contractually guaranteed. This is different from ChatGPT — API usage has stricter data protections. You can verify this directly in OpenAI's usage policies at platform.openai.com.

AI can be wrong.

AI-generated content in Gunimi is a starting point, not an authoritative record. Always verify critical business information against primary sources. The responsibility for decisions made using AI-assisted information remains with the human making that decision.