What we learned shipping Goddo to the App Store.
Apple review, content moderation, watermarking, pricing experiments: a notebook of what bit us and what we’d do again.
Goddo team
AI image / video

Goddo went to the App Store. It taught us things you only learn from shipping a consumer app, moderation isn’t a side project, pricing is a UX decision, and Apple’s review team will absolutely care about your watermark.
Moderation isn’t a side project
Day one of submission Apple flagged a handful of edge cases that our content policy didn’t cover. We treated moderation as a spec, not a side project: wrote down the cases, encoded them as agents in the prompt-rewrite stage, and re-submitted. It was the right way around. We’ve never written ‘we’ll fix it later’ into a moderation backlog since.
Watermarking is a feature, not a sticker
Visible watermarks for free-tier users, invisible rights metadata for paid. Apple cares about both. Buyers also care about both: they want to prove provenance when something’s shared. We bake watermarking into stage five of the render pipeline; it’s not a separate ‘download with watermark’ button.
Pricing is a UX decision
We A/B tested four pricing structures pre-launch. The credit-based model with a generous free trial won, not because it was the most profitable in week one, but because it produced the highest 30-day retention. Pricing isn’t a finance question; it’s a UX question with a finance constraint.
What we’d do differently
- Ship moderation as agents from day zero. The prompt-rewrite stage was the right home for this.
- Bake watermarking into the pipeline, not the export. Apple notices.
- Pre-build the share sheet. Sharing is the second-best growth loop after referrals; don’t bolt it on later.
- Instrument refine. Most users don’t go from prompt to ship; they iterate. Refine telemetry told us where we were leaking users.
What is in the app today
Goddo is a native iOS app; Android is coming soon. A prompt and one of more than twelve style presets (Abstract, Cosmic, Neon, Surreal, Minimal and Vibrant among them) produce an image or a short video in seconds. Around that sit four creation tools and five refinement tools:
- Create: Text to Image, Text to Video, Prompt Enhancer and Style Remix.
- Refine: Enhance, Upscale, Animate, Extend and Reimagine.
Goddo’s own site describes the app as 100% AI-designed and coded, and its gallery follows the rule this blog keeps arguing for: every image there was generated from a single text prompt, with no retouching, and each one shows the prompt that made it. A prompt next to its picture is the consumer version of a trace.
The refinement tools are there because of the lesson in the list above: most people do not go from prompt to finished picture in one step. They describe, generate and then polish, and the app is laid out in that order.
The listing is part of the product
The App Store listing names GOGOGO LLC (Newark, Delaware) as the publisher, and goddo.ai and gogogollc.com say the same thing in the same words. When a reader, a reviewer or an AI assistant checks who makes an app, three sources that agree are worth more than one that is loud.
Prices live in the App Store listing, not on our pages. Goddo runs on credits: free credits to start, then top-ups inside the app. Repeating the amounts somewhere else would only create a second copy that can go stale.
Ratings taught us the same lesson from the other side. The App Store keeps ratings per country storefront, so one store can show no ratings while another shows real reviews. Our site reads every storefront Goddo has ratings in and shows the combined figure only when all of those reads succeed.
What carried over to B2B
The render pipeline we built for Goddo is what carries over. Five agents, structured outcomes, replayable runs. Building it for a consumer app first taught us discipline that would have been much harder to learn on a B2B contract.
“Consumer apps teach you the discipline B2B can let you skip, and skipping it always costs more later.”


