How PerfIQ uses AI
PerfIQ uses Anthropic's Claude models, called through Anthropic's API with PerfIQ's own key.
- What is sent
- What PerfIQ does with it
- What Anthropic says about its own handling
- Learning across companies
- What AI answers can and cannot do
- What we do not do yet
What is sent
To answer you, PerfIQ sends the relevant parts of your information to Anthropic: your messages to Q, your brief and answers, figures you uploaded, text from PDFs you attach, the notes Q keeps about your company where that feature is on, and the analysis built from them.
Some research steps use the web search tool offered through Anthropic's API; the search words are written by the model and can include your company name, industry or competitor names.
If you give a website address, PerfIQ reads that public page itself and keeps what it reads as a note on your company.
What PerfIQ does with it
PerfIQ does not train or fine-tune AI models on your content.
PerfIQ keeps your conversations, notes and results so you can come back to them.
PDFs you attach are deleted after 30 days, when you replace them, or when you delete them; the notes made from them stay until you remove them.
What Anthropic says about its own handling
These are Anthropic's statements, dated, and not our guarantee. Read them at the source.
As of 2026-10-30, Anthropic states that it does not use inputs or outputs from its commercial products, including the API, to train its models (source).
As of 2026-10-30, Anthropic states that API inputs and outputs are automatically deleted from its back end within 30 days of receipt or generation, except where longer retention is required by law or to enforce its usage policy, and that content flagged by its safety systems can be kept longer (source).
Learning across companies
PerfIQ also learns from patterns across companies, such as industry, size band, the pattern found, scores and ranges.
Demo, sample, test, staff and internal accounts are left out.
This is on by default; to opt out, write to legal@theperfiq.com.
We call this de-identified and aggregated, not anonymous.
If you delete your account, the learning records tied to it are deleted too.
What AI answers can and cannot do
Q's answers are generated by AI and can be incomplete, out of date or wrong, and the Terms of Service ask you to review results before you rely on them.
Chat messages, questions and exposure scans have per-account limits, and the service has a daily limit on AI usage.
What we do not do yet
- We do not yet have full protection against instructions hidden inside a document or web page that try to steer Q.
- We do not have a zero-data-retention agreement with Anthropic.
- You cannot yet opt out of cross-company learning from inside the product; you write to us.