Reference Studies & Attachments

Reference Studies and Attachments let you give the AI context it didn't have before — your own documents, your own research history, anything that should shape what it does next.

What you can attach

You can upload files into a session: PDFs, Word documents, spreadsheets, images. The AI reads them and uses them as context for whatever research follows.

Common uses:

• A product spec the AI should keep in mind during interviews.
• A competitor's pricing page you want discussed.
• Internal research from your team that should anchor a new study.
• Customer interview transcripts you want the AI to learn from.

Files stay attached for the duration of the session. The AI references them when generating plans, asking questions, and producing reports.

Reference studies

Beyond raw attachments, you can pull in past research sessions as references. If your team did a study six months ago on the same audience, you can link that session as a reference and the new study can build on it — re-use personas, continue where it left off, or compare new findings against old ones.

This is the continuity mechanism for research that spans months or years. Each new study doesn't start from zero; it can lean on what came before.

How attachments show up in practice

When you upload a document, it becomes available to the AI as numbered context — something like [#1 pricing-analysis.pdf] — that the AI can quote from, summarize, or reference during the conversation.

You don't need to pre-process files. PDFs are read, tables are parsed, images are described. The AI handles the structure.

When to use attachments

Use attachments when:

• The research needs to be grounded in your specific product, market, or customer.
• You have internal context the AI couldn't know — your company's strategy, your team's prior findings.
• You're validating an external framework against your own data.
• You want the AI to compare new findings to known baseline.

Skip attachments when:

• The research is exploratory and broad. Adding your company's strategy doc to "understand Gen Z" just narrows the AI unnecessarily.
• The attachments would overwhelm the conversation with too much context. The AI's working context is finite — too many attachments means the AI pays less attention to each.
• The information is sensitive and shouldn't be passed to AI systems at all.

Limits

• File size and type matter. Very large PDFs may take time to process.
• The AI reads what it can parse cleanly. Scanned documents without OCR, password-protected files, or unusual formats may not be fully usable.
• Attachments are scoped to the session. They don't persist across sessions automatically — you re-upload when you start a new session, unless you reference a past study.
• Reference studies are read-only. The AI can use them as context, but can't modify them.

Privacy

Attachments and references are scoped to your account or team. Other users can't see them, and they're not used to train external models.

For sensitive documents (customer PII, internal financials), follow your organization's data-handling policies before uploading.