AI Disclosure
Phloem is AI-assisted by design. This disclosure explains what that means in practice and where the limits are.
1. Where AI is used
AI models help research companies from permitted sources, summarise findings, score commercial fit, explain why a company could become a customer, and draft personalised messages.
Each AI-assisted result is presented with the sources behind it so you can verify it before acting.
2. Human control
Agents run in suggest, approve or auto mode, within the budget you set per mission and per agent.
Outbound communication passes a compliance evaluation and, where your policy requires it, a human approval step before sending.
AI does not make decisions about individuals that produce legal or similarly significant effects on them.
3. Limits of AI output
Generated text and scores can be incomplete, outdated or wrong. Treat them as drafts and research aids, not as verified facts or advice.
Phloem does not guarantee commercial results, and generated messages are not automatically lawful in every jurisdiction.
4. Data and model providers
Prompts and the content needed to produce a result are sent to model providers acting as sub-processors. Sensitive fields are masked before they leave the platform where classification identifies them.
Workspace content is not used to train foundation models.
Legal, privacy and security contact runs through the Phloem support form and privacy centre — we do not publish individual inboxes.