The model is the easy part. What building Dantiva taught me

Description and use

Dantiva built a paid AI video service using Google models, implementing a token ledger, idempotent jobs, and a unified experience across web and Telegram. The product launched subscriptions and plans to add a second provider.

Problem

Building a paid AI video generation product that integrates Google models, handles payments, user accounts, and token ledger while ensuring compliance and consistent UX across web, Telegram bot, and Mini App.

Outcome

Dantiva launched a paid AI video service with a robust token ledger, idempotent job handling, and unified user experience across web and Telegram, achieving early revenue and setting the stage for multi‑provider expansion.

Workflow

• 1. User signs up via Google or email on web, or via Telegram ID/mini app payload. 2. User selects template and uploads photo. 3. Prompt rewritten by Gemini model. 4. Video generation request sent to Veo 3.1 (Fast/Lite) via Cloud Run. 5. Ledger reserves tokens, tracks job ID, ensures idempotency. 6. On completion, tokens captured; on failure, refunded. 7. Result stored in library, displayed across all interfaces. 8. Admin panel handles support, refunds, and legal center. 9. Subscriptions managed manually, with plans and top‑ups. 10. Future: add second provider, automatic renewal, interview users.

Evidence

Generation costs real money on every call, and a call can fail halfway. So the balance is not a number you subtract from. It is a ledger: reserve tokens when a job starts, capture them when the video arrives, refund them if the provider fails.

Source

All the templates start from a photo of a person. Google Veo blocks image to video with recognizable people.

Source