Frequently asked questions

An agentic creative canvas: you brief the AI director once, it picks the right model across image, video, lip-sync, and audio, dispatches the job, and drops the result back into the canvas. 15 models, 4 providers (fal, Replicate, Luma, Hedra), one workflow.

No. Type or paste a brief in plain English ("Hero shot of a ceramic mug, golden hour, 1:1"). The director chooses the model and parameters. Pull the result onto the canvas to compose.

Product photography, ad creatives in any aspect ratio, brand-consistent images via your Brand Kit, short cinematic videos, spokesperson lip-sync, and longer multi-scene storyboards through the Director.

Three monthly plans — Plus ($19), Pro ($49), Ultra ($149). Each comes with a monthly credit allowance that resets every billing cycle. Credits are consumed per generation: image (1), video (5), lip-sync (8), brief (2).

Credits are automatically refunded when the provider returns an error. You only pay for renders that actually start successfully.

Yes. Go to Settings and paste your Luma, Hedra, or OpenAI keys. When a personal key is configured, the system uses it instead of ours — useful if you want to track spend directly on a provider account.

No. Each renewal sets your balance to the plan's monthly allowance. Use them or lose them.

Yes. Cancel from Settings → Subscription. You'll keep access until the end of the paid period, then downgrade to Free.

Photon Flash / Photon 1 for images, Ray Flash 2 / Ray 2 for video, Kling v2 / v1.6 / Hunyuan via fal, and Hedra Character-3 for spokesperson lip-sync. The director scores models on style match, quality, speed, resolution, and cost — then dispatches.

A set of brand assets (logo, primary/secondary color, voice tone, character reference photo) that get injected into every generation prompt. Set one up in Settings → Brand Kit. The director references it on every render so visuals stay on-brand.

Provider render times vary: image ≈ 8–15s, video ≈ 30–120s, lip-sync ≈ 60–90s. The canvas chat shows a live progress card per active job; the video drops in inline when the provider finishes.

Your account profile (from Clerk), encrypted BYOK API keys (AES-256-GCM, per-user IV), saved canvas snapshots, brand kits, and generation history for credit auditing. Generated outputs are stored on Vercel Blob under your user-scoped prefix.

No prompts are used to train Best AI Marketing Software's systems. Provider terms vary — fal, Replicate, Luma, and Hedra each publish their own data-use policies. BYOK gives you the strongest control: the prompt goes directly to your account.

Every user-authored prompt is sanitized to strip invisible Unicode (tag block, zero-width, bidi overrides) before it reaches the LLM. Brand Kit free-text is sanitized at write and read time.

Go to Settings → Subscription → Manage. You'll be sent to the Stripe Customer Portal where you can upgrade, downgrade, or cancel.

Sometimes Stripe webhook events take a minute to process. Refresh after ~60s. If it persists, contact support with your Stripe receipt and we'll reconcile manually.

We honor refunds for full-month windows where no renders were performed. Contact support within 7 days of the charge.