VDF AI Data

Getting started with VDF AI Data

Upload your first file, connect your first app, and run your first grounded query — all in under fifteen minutes.

3 min read
On this page

What you need before you start

Three small things:

  1. Access to your VDF AI workspace. If you haven’t signed in yet, start with the Getting Started guide.
  2. One real document. A policy, a contract, a report, a transcript — something you’ve referenced more than once in the past month.
  3. (Optional) one connected-app credential. If you want to wire in Google, Microsoft, Confluence, or another platform — have your login ready.

Start small. Don't try to bring your whole company knowledge base in on day one. One file and one connected folder is enough to feel the difference. Expand from there.

Step 1 — Upload your first file

From your workspace, open the Data area (sometimes called Knowledge or Sources). Click the upload button and drop in your document.

Supported formats include:

  • Text documents — PDF, DOCX, Markdown, plain text
  • Spreadsheets — XLSX, CSV
  • Slides — PPTX
  • Transcripts — text-based meeting transcripts (.txt, .vtt, .srt)

Some platforms also support images, scanned PDFs (with OCR), and audio transcripts — check your workspace’s specific capabilities.

What to expect

After upload:

  • VDF AI processes the file — usually in seconds for small docs, a minute or two for large ones.
  • The file appears in your Data list with a “ready” indicator.
  • It’s now usable from any conversation, agent, or network in your workspace.

Visibility defaults vary by workspace. Some workspaces upload as private by default; others share with the team. Check the visibility setting after upload if you're unsure.

Step 2 — Ask your first grounded question

Open Chat. In the prompt, reference the file:

“Using the attached [filename], pull out every commitment with a date and turn it into a checklist.”

Or more simply:

“Summarize [filename] in five bullets.”

The response will cite which sections of the file each part of the answer came from. Click through to verify — that’s the loop you’ll do every time.

Step 3 — Connect your first app

Connected apps remove the upload step entirely. Once an app is connected, VDF AI can reference content from it on demand.

Go to the Connections area in your workspace. Pick an app to connect — Google, Microsoft, Confluence, GitHub, Jira, Slack, Zoom, GitBook, or others available in your workspace.

What happens during connection

  1. You’ll be redirected to the app’s login screen.
  2. The app shows you what VDF AI is asking permission to access (a folder, a workspace, a space, a project).
  3. You approve — or scope down to a specific area.
  4. You’re redirected back to VDF AI, and the connection appears as “active.”

Scope tightly the first time. Don't grant access to your entire Google Drive on day one. Connect a specific folder. You can expand later — narrowing is harder than starting tight.

Step 4 — Reference connected content in a conversation

With a connection live, open Chat and reference content from the connected app:

“Find all my Confluence pages tagged ‘onboarding’ and summarize the recurring themes.”

“What did our team write in our ‘Q3 launches’ Confluence space about the migration?”

“Across the open Jira tickets in project ALPHA, what’s the most common type of issue?”

The output will reference specific pages, tickets, or documents — clickable links back to the source.

Step 5 — Combine sources in one question

The most powerful pattern is combining sources:

“Compare the renewal terms in the attached contract with what our standard playbook (in the connected Confluence space) says about renewal.”

VDF AI reads both — your uploaded file and the connected Confluence space — and produces a comparison with citations on both sides.

This is the kind of work that’s genuinely hard to do by hand and trivially easy once Data is set up.

Step 6 — Review what’s in your Data area

After a week of use, open your Data area and take stock:

  • Files you’ve uploaded — what’s still useful? Anything stale or one-off you can remove?
  • Connected apps — are the connections still active? Have permissions changed?
  • Searches you’ve made — what kinds of queries have worked well?

A 10-minute weekly review keeps your Data area sharp. A stale Data area produces stale answers.

Common first-week patterns

The teams that succeed with Data tend to do a few of these in their first week:

Upload one important reference doc

A policy, a contract, an onboarding guide — something you'd otherwise re-read often.

Connect one team workspace

Confluence space, Google Drive folder, GitHub repo — wherever your team's writing lives.

Ask one question across them both

The "aha" moment is usually when you ask something that requires both sources to answer.

Generate one structured output

Turn a long doc into a checklist or summary — and verify the citations match the source.

Troubleshooting your first run

The file uploaded but the answer doesn’t reference it. Name the file explicitly in your prompt. “Use the attached MSA-2025.pdf” works better than “use my file.”

A connected app shows as “needs reauthorization.” The connection’s permissions expired or the app was disconnected on the source side. Reconnect from the Connections area.

The answer feels generic. You may be asking too broadly. Narrow to a specific folder, file, or page in your prompt. “Use only the pages in the ‘Engineering Standards’ space” gives sharper answers than “use Confluence.”

The output cited the wrong page. Two possibilities — the source content actually overlaps (multiple pages cover similar ground), or the connection is stale. Re-ask the question with a narrower source scope and verify against your sources directly.

What to read after your first day