We turn your call transcripts into content gold. Here's a transparent look at what happens to your data at every step — no fine print, no surprises.
From transcript to content — every step, explained
Only the text. Never audio. Never video. Never screen recordings.
We send the transcript text to Claude (Anthropic) and OpenAI via their APIs. API usage is never used to train their models — this is contractually guaranteed by both providers.
Insights, frameworks, and content ideas are saved to your account. Row-level security means no other user — not even other MagnetLab customers — can access your data.
Edit, use, or delete. When you delete a transcript, all extracted data is permanently removed. No backups retained.
Six promises we make to every customer
Your transcripts, insights, and content are never used to train any AI model — ours or anyone else's. Claude and OpenAI's API policies contractually prohibit using API data for training.
You own everything you put into MagnetLab and everything it produces. We claim no rights to your content, insights, or intellectual property.
Row-level security enforces strict data isolation at the database level. Your account is a vault — no other MagnetLab user can access it, period.
When you delete a transcript, all extracted knowledge, content ideas, and embeddings are permanently removed via cascading deletes. There is no "soft delete" or hidden retention.
All data is encrypted via TLS in transit and AES-256 at rest through our infrastructure provider (Supabase on AWS). API calls to AI providers use encrypted connections.
Anthropic (Claude) and OpenAI process your data via API and do not retain it after the response is returned. Zero data retention is the default for API customers.
Two ways to use the content pipeline — pick what works for you
Don't connect anything. Copy and paste specific transcript sections you're comfortable sharing. Same as using any other AI tool.
Connect Grain, Fireflies, Fathom, or any recorder via webhook. Transcripts flow in automatically and are processed within minutes.
Your call recorder already has your transcripts. Most of them train their models on your conversations and give you nothing in return except a searchable archive.
MagnetLab is the opposite — we extract genuine value from that data (insights, content, posts in your voice) while giving you stronger privacy protections than the tools already sitting in your meetings.
The question isn't whether AI touches your call data.
It's whether you're getting anything back for it.
We use API access, not consumer products — here's why that matters
Knowledge extraction & content generation
Search embeddings only (not full transcripts)
Important distinction: these protections apply to API access (what MagnetLab uses), not to consumer products like ChatGPT or Claude.ai where different policies apply.
No. The auto-connect feature is completely optional. You can paste transcripts manually, upload files, or skip the content pipeline entirely and just use MagnetLab for lead magnets and funnels.
No. We use their APIs, not their consumer products. Both Anthropic and OpenAI contractually guarantee that API data is not used for model training. This is a key distinction from using ChatGPT or Claude.ai directly — the API has stronger privacy protections.
Most call recording platforms do use your data for model improvement, often with vague opt-out policies. MagnetLab only receives the text transcript — we never access the audio, and we give you actual value from data you're already sharing with those platforms.
Yes. Delete any transcript and all associated knowledge entries, content ideas, and vector embeddings are permanently removed via cascading database deletes. You can also delete your entire account.
Only you (and your team members, if you've set up a team). Row-level security policies enforce this at the database level — it's not just application logic, it's a database-level guarantee.
The text of your transcript (truncated to 25,000 characters) is sent to Claude for knowledge extraction and content ideation. Short summaries of extracted insights are sent to OpenAI for generating search embeddings. The full audio or video recording is never sent anywhere.
Not yet. We're a growing platform and SOC 2 Type II is on our roadmap. In the meantime, we build on enterprise-grade infrastructure (Supabase on AWS, Vercel) that maintains their own SOC 2 and ISO certifications.
Absolutely. The content pipeline is one feature of many. You can create lead magnets, build funnels, capture leads, and run email sequences without ever connecting a call recorder or pasting a transcript.
Start with lead magnets, add the content pipeline when you're ready. Your data stays yours every step of the way.
Create Free Account