Prompt Orange
A guided AI prompt builder that scores your prompts out of 100 and tells you exactly how to improve them.
Why this project. Why now.
Most people using AI tools get mediocre results — not because the AI isn't capable, but because they don't know how to ask well. The founder wanted to build a tool that closed that gap: something that met users where they were, asked the right follow-up questions, and gave them a prompt that actually worked. Free to start, no signup required.
Where the complexity lived.
The challenge was making prompt engineering feel effortless for someone who has never heard the term. The wizard had to feel like a conversation, not a form. The scoring had to be specific enough to be useful, not just a vanity number. And it needed to work across four very different use cases — code, images, data, and writing — each with its own set of expert questions and output criteria. A companion iOS app with StoreKit credit purchases had to ship alongside the web product.
The core product, in four capabilities.
Describe your goal in plain English. The wizard classifies your intent, picks the right category, and asks smart follow-up questions to fill the gaps the AI needs.
Every generated prompt is scored out of 100 with a criterion-by-criterion breakdown — so you know exactly what's working and what to fix.
Use credits to iterate. Answer follow-up refinement questions and get an improved prompt until the score — and the output — is exactly right.
The full wizard experience as a native SwiftUI app, with StoreKit in-app credit purchases, prompt history, and category browsing.
Seven days. Every one accounted for.
We run a structured sprint across seven days. Each phase is deliberate — no wasted motion, no scope creep, no surprises at the end.
Designed the wizard config system and category schemas for Code, Image, Data, and Writing — each with its own expert question set, output preference, and scoring criteria. Set up Next.js + DynamoDB infrastructure.
Built the intent capture flow, AI-powered category classification, and the dynamic question renderer. Each category surfaces a different set of smart follow-ups based on the user's goal.
Integrated the generation pipeline producing structured, expert-framed prompts. Scoring engine evaluates each prompt across multiple criteria and returns a score out of 100 with a breakdown.
Credit wallet system with Stripe checkout on web. Refinement flow that consumes credits and returns an improved prompt with updated score.
Native SwiftUI app mirroring the web wizard. StoreKit integration for in-app credit purchases. Prompt history view and category browsing from the home screen.
Browsable prompt template library with pSEO pages per category and role. Business tier landing page. Production deployment on Vercel with Tech Due Diligence documentation.
What happened after we shipped.
Prompt Orange launched simultaneously on web and iOS — the same credit system powering both. Users could build, score, and refine prompts for code generation, image creation, data analysis, and writing from day one. The pSEO template library added long-tail organic reach from launch.
Your product could be next.
We take one project at a time — full attention, senior-level build, delivered in seven days. If you have an idea worth shipping, let’s talk.
The £299 scoping session is credited in full if you proceed. We confirm fit within 24 hours.