Drive and documents

Once Google Drive is connected (part of Connect Google Workspace), your AI gets a new superpower: it can search your existing documents, read them, quote from them, and write back to them on request.

This chapter is about using that well.

Search and retrieval

The simplest pattern: ask, in plain language, about documents.

“Find the spec doc for the redesign.”
“Show me the contract draft I sent Anna last month.”
“Pull up the meeting notes from the kickoff with Mike.”

Your AI searches your connected Drive, returns the top matches, and offers to open or summarize them. If there are multiple plausible matches, it lists them and asks.

Search uses both filename and full-text content. “The doc where we talked about the dishwasher” works even if the filename doesn’t mention dishwashers.

Summarizing

“Summarize the redesign spec in three paragraphs for someone who hasn’t read it.”

“What are the open questions in the contract?”

“Tell me what the meeting notes say about next steps.”

Your AI fetches the document, reads it, and produces an artifact with the requested summary. The original document is unchanged; the summary is a new artifact you can keep, share, or revise.

For long documents, you can scope: “Summarize just the section about pricing.” Section-aware summaries are fast and surprisingly accurate.

Quoting back

When you want exact phrasing, ask for quotes:

“Quote what the contract says about termination.”

“What’s the exact language about working hours in the offer letter?”

You get the relevant passage verbatim, with a citation to the source document. This matters for legal and contractual work where paraphrasing isn’t safe.

Cross-document work

The interesting use cases involve more than one document:

“Compare the two contractor contracts and tell me where they differ on scope, payment terms, and timeline.”
“I have three project briefs in the redesign folder. Synthesize them into a single one-pager.”
“Find every place across my docs where I’ve quoted the same statistic. Are they consistent?”

Your AI does this by reading each relevant document and producing the comparison or synthesis as a new artifact.

The cost (in AI usage) of reading many documents at once isn’t trivial, so for very broad asks (“synthesize every doc in this folder”), you’ll be asked to confirm scope.

Writing to documents

Your AI can also write back, with approval.

“Update the project brief: add a new section on the dishwasher decision.”

“Add my decision notes from yesterday’s call to the kickoff doc.”

The write-back is gated by default. You see the proposed change as a diff (added paragraphs highlighted, edits inline), approve, and the document is updated.

Writing back to a shared document means collaborators see the change. Your AI doesn’t try to be sneaky about it; the edit appears in Google Docs’ history with the appropriate attribution.

Using docs as project source

A pattern that pays off: when you start a project, tell your AI which documents are the source-of-truth references.

“On the redesign project, the canonical spec is the doc titled ‘Redesign v2 spec’. The brief in our shared folder is the kickoff brief. The contract is in the legal folder under Acme.”

Your AI links these documents to the project’s record. When you (or anyone on your instance) asks something about the redesign, your AI knows which documents to consult first.

This is also where projects get useful for teams. Everyone working in that project’s chat gets the same context, without having to remember which doc is current.

When to draft in a doc vs. in an artifact

A subtle but useful distinction:

Drafts that need to be in your Drive for sharing with people outside Jootle: write directly to a doc, or produce an artifact and export to a doc.
Drafts that live inside your work in Jootle: produce as artifacts (see Artifacts). They’re versioned, attached to entities, and don’t clutter your Drive.

A rule of thumb: if the doc is going to be sent to or edited by someone who doesn’t use Jootle, write to Drive. If it’s internal to your Jootle work, keep it as an artifact.

Permissions and scope

When you connect Drive, your AI gets access to documents you specifically authorize at OAuth time. By default, that’s the scope you pick. Most customers grant access to a specific Drive folder rather than everything.

Inside that scope, your AI can:

Read documents you can read.
Write to documents you can write to.
Not access documents outside the scope (including ones you have access to but haven’t included).

You can revoke access at any time from your Google account’s security settings, or disconnect the connection inside Jootle. Disconnection halts all read/write immediately.

A small note on file types

Your AI handles common file types well:

Google Docs, Sheets, Slides natively.
PDFs (read-only, including OCR for scanned).
Microsoft Office (.docx, .xlsx, .pptx) for reading.
Plain text and Markdown natively.
Images with text are OCR’d.

Files in formats your AI doesn’t handle (CAD files, video, exotic formats) can be referenced and linked, but content extraction is limited.