Frequently Asked Questions

What's the difference between "Connected" and "Voice Trained"?

Connected means BabbitAI has permission (via that platform's login) to post on your behalf and, where allowed, read your existing posts. Voice trained means BabbitAI has actually learned how you write for that platform — either by reading enough of your past posts, or from writing samples you pasted in yourself.

You need both to get on-voice content for a platform. Being connected without a trained voice means BabbitAI can post for you, but won't yet sound like you on that platform specifically — so every "Connect" on BabbitAI now walks you straight into training, rather than leaving that step for later.

Why does LinkedIn always ask for 3 writing samples?

LinkedIn's API doesn't grant regular apps — including BabbitAI — permission to read a member's existing posts. That access exists only as a restricted, partner-only permission LinkedIn grants case by case, and BabbitAI doesn't have it. This isn't a missing feature we're building toward; it's a wall LinkedIn puts up for every app in our position.

So for LinkedIn specifically, pasting 3 of your own posts is the only way to train an authentic LinkedIn voice — there is no auto-read alternative, now or later, unless LinkedIn changes that policy.

Why did Instagram or TikTok ask me for 3 samples instead of reading my posts?

Unlike LinkedIn, Instagram and TikTok do let BabbitAI read your recent posts once you accept the "read my posts" prompt during connect — this is the normal path and works for most accounts. You'll be asked for manual samples instead only when that automatic read comes up short, for example:

• Not enough recent posts — we need a handful of usable posts with actual text/captions to learn from; a brand-new or very quiet account may not have enough yet.
• Account type isn't supported for reading — some post-reading endpoints require a Business or Creator account; a personal account can still connect and post, but voice-reading may not be available.
• A temporary read failure — the platform's API didn't respond as expected; retrying later sometimes resolves this, but samples work immediately.

In any of these cases, a popup explains which of the above happened and takes you straight to the manual-samples section so you're never left "connected" without a trained voice.

What happens if I add samples manually — do they get tagged to the right platform?

Yes. The "Which platform is this voice for?" selector on the Train New Voice section tags whichever platform you pick (or leaves it General if you're training a voice that isn't platform-specific). That tag is what BabbitAI uses to know a platform has a trained voice, so pick the platform honestly rather than leaving it on General if the samples are meant for a specific one.

Can I retrain or add another voice later?

Yes — you can train as many voice profiles as your plan allows (see your plan's brand-voice limit on the Billing page), and you can delete any existing profile from the Voice Training page to make room for a new one.

Do you keep my photos and videos?

Depends which upload you mean — BabbitAI has two separate photo/video features (see the next question for the size/browser details):

• Uploaded for caption ideas: processed in memory only — never permanently stored. The same goes for audio from voice memos or a video's soundtrack, sent to Groq's Whisper API for transcription and never saved.
• Attached to actually post to Instagram or TikTok: stored in cloud storage so those platforms can fetch it, retained while your account stays active.

See our Privacy Policy for the full breakdown of what we store and for how long.

How do I get a refund?

Payments are non-refundable. Cancel anytime from your Billing settings — access continues through the end of your current billing period, but cancel at least 1 day before your renewal date, or you'll be charged for the next cycle before the cancellation takes effect.

The only exceptions are a duplicate or mistaken charge, or a charge that went through after your cancellation should have already applied — reach out through the Support link in the app and we'll fix it. If an outage or a failed generation on our end keeps you from using BabbitAI, you'll get a credit on your next bill instead of a cash refund: 25% if it's resolved within 5 business days, 50% if it takes longer.

How do I contact sales about Agency or Enterprise plans?

This is specifically for Agency/Enterprise plan questions — pricing, seats, custom terms, and the like. Email admin@babbitai.com directly for that. If your question is a general support issue instead, use the search above or the Support link in the app — sales isn't the right inbox for those and it'll get you an answer faster.

What are the size/browser requirements for photo & video uploads?

There are two separate photo/video features on the Generate page, with different limits — don't confuse them:

• Uploading a photo/video for caption ideas (Creator plan+): JPG/PNG/WEBP photos up to 20MB, or MP4/MOV/WEBM videos of any length — BabbitAI analyses one representative frame, so video length and resolution don't matter here.
• Attaching media to actually post to Instagram or TikTok: MP4/MOV/JPEG/PNG, up to 50MB. This is a separate attachment step from the caption-ideas upload above.

For video specifically, BabbitAI can optionally read the first 90 seconds of a clip's spoken audio (via Whisper transcription) to write a caption or script that references what's actually said, not just what's visible. This only works in Chrome, Edge, or Firefox on desktop, or any browser on Android — it does notcurrently work in Safari (desktop) or in any browser on iPhone/iPad, since Apple requires every iOS browser to use Safari's engine underneath. On unsupported browsers, or if a video has no audio, captions still generate fine from the visual frame alone — you just won't get the audio-aware boost.