Documentation

Findery is a search-first knowledge layer over the content your business already has spread across SaaS tools. This page covers what Findery does, how to get started, and how each integration works · in particular the Zoom Cloud Recordings integration.

What Findery does

You sign up. You connect the tools your business already uses (Help Scout, Circle, ClickUp, YouTube, Google Drive, Loom, Zoom). We pull the content, generate embeddings, transcribe audio + video, and serve it back via:

• Search · plain-English search across every doc, video, lesson, and ticket; results show the snippet that matched + a deep-link into the source.
• Ask · a question gets a cited answer drawn only from your own content (no external web data, no general LLM knowledge).
• SOPify · an AI SOP-builder that drafts Standard Operating Procedures from your existing resources and shows you exactly which lesson, doc, or video each line came from.

Every customer gets their own subdomain ({slug}.findery.io) or can point a custom domain. Each customer’s data lives in a dedicated database (Neon Postgres) and dedicated object storage buckets (Cloudflare R2), isolated from every other tenant’s.

Getting started

1. Sign up at findery.io. 60-second form; no credit card.
2. Verify the email link. We provision your workspace · a private database, storage buckets, and a {slug}.findery.io subdomain.
3. Connect a source on the Integrations page · paste an API key or click through an OAuth flow (depending on the source).
4. Wait a few minutes while we ingest. Search and Ask light up automatically as content lands.

The free trial caps ingest at 5 resources per source. Subscribe to lift the cap.

Integrations

Zoom integration

The Zoom integration uses OAuth 2.0 with the user-managed install flow. When you click Connect Zoom on the Integrations page, you’re redirected to Zoom’s consent screen and asked to grant Findery these specific scopes:

• cloud_recording:read:list_user_recordings · so we can list your past cloud recordings
• cloud_recording:read:list_recording_files · so we can download the audio file for each recording you pick
• cloud_recording:read:meeting_transcript · so we can fetch Zoom’s auto-generated VTT transcript when one is available
• user:read:user · so we can identify which Zoom account is connected

We do not request admin scopes. Findery only ever sees recordings that belong to the Zoom user who authorized the connection · not other people in your Zoom workspace.

What gets imported

After you connect, Findery shows you the past 90 days of your cloud recordings in a picker · meeting topic, date, duration, file size, and whether Zoom generated a transcript for it. You check the boxes for the meetings you want indexed. For each one:

1. We download the best audio file (M4A preferred, MP4 fallback).
2. We upload it to your tenant’s R2 audio bucket.
3. If Zoom generated a VTT transcript, we parse it into timestamped segments and store it. Done.
4. If there’s no Zoom transcript, we queue the recording for our own transcription pipeline (faster-whisper on a dedicated GPU). Transcripts typically land within 5–15 minutes per hour of audio.

After ingest, the recording is searchable like any other resource. The search page has a Zoom filter pill; results that came from a Zoom recording show “from Audio” and deep-link to the specific timestamp inside the audio player on the resource page.

What we don’t do

• We don’t pull recordings you didn’t pick. Every ingest is explicit.
• We don’t share your Zoom data with any third party. Audio + transcripts live in your dedicated R2 buckets and your dedicated Postgres database.
• We don’t make new recordings or join meetings. The integration is one-way read-only.
• We don’t modify or delete anything on the Zoom side. Even if you delete a recording inside Findery, the original stays in your Zoom account.

Disconnecting

You can revoke Findery’s access at any time:

• From Zoom: sign in to Zoom Marketplace, go to Manage → Installed Apps, find Findery, and click Uninstall. This revokes the OAuth token and Findery’s API calls start failing immediately.
• From Findery: on the Integrations page, click Reconnect on the Zoom card to swap the authorization, or contact support to wipe the stored tokens entirely.

Already-ingested recordings stay in your Findery workspace after you disconnect · they’re your data. To remove them, delete them in the Findery UI or ask support to wipe them.

Data handling

• Where your data lives: Postgres on Neon (US-East), R2 on Cloudflare (global), application servers in the US. Per-tenant isolation at every layer.
• Encryption: TLS in transit; AES-128 (Fernet) at rest for all API keys and OAuth tokens; bcrypt-12 for passwords; Cloudflare R2 server- side encryption for object storage.
• What we never do: sell, rent, license, or train external AI models on your content. Period. See the Privacy Policy for the full breakdown.

Need help?

The Support page has email + in-app paths. Most questions get a same-day answer.