AI-powered mobile app
A mobile app with AI features such as chat, image recognition or smart search, built with costs and privacy under control.
GradaxEuropean companyOrder outright
Founded 2014·Built and operated in Europe·2 to 14 weeks
About this service
We build iOS and Android apps that use AI models for a defined job: an assistant that answers questions from your content, image recognition for products or documents, transcription, recommendations or smart search. The AI calls run through your own server, so API keys stay hidden and usage costs can be capped. We also design for when the model is wrong.
Pick one job the AI does well
The weakest AI apps add a general chat box and hope users find a use for it. The strongest give the AI one clear job tied to the product: answer questions about this company's policies, identify this type of part from a photo, summarise this kind of document, suggest what to cook from these ingredients. A defined job can be tested, measured and improved; a vague one cannot.
We start by writing that job down with examples of good and bad answers, then build a quick prototype and test it on your real examples before building the full app. Sometimes the test shows the AI is not reliable enough for the job yet, and it is far better to learn that in two weeks than after a full build.
Costs, keys and wrong answers
Three practical problems catch out many AI app projects. First, API keys built into the app can be extracted by anyone who downloads it, and then run up bills on your account. We route all AI calls through your own small server, which keeps keys hidden and lets us apply limits per user.
Second, usage costs scale with users. A feature that costs a fraction of a cent per question can become a real monthly bill at volume. We estimate cost per active user up front, set caps and show usage on a dashboard. Third, models are sometimes confidently wrong. Where accuracy matters, answers are drawn from your own documents and shown with their sources, and users can flag bad answers so they can be reviewed and fixed.
- API keys on the server, never in the app
- Per-user limits and a monthly cost cap
- Answers grounded in your content, with sources shown
- A feedback button that logs weak answers for review
What you get
Everything below is in the price. Nothing here is an upgrade.
- Feature specification with test cases
- Mobile app for iOS and Android
- Backend that handles AI calls, limits and logging
- Evaluation results on your test examples
- Source code and documentation
What changes
- One clear AI feature that does a defined job, tested on real examples
- API keys kept on a server, never inside the app
- Usage limits and cost monitoring so a busy day does not bring a surprise bill
- Answers grounded in your own content where accuracy matters, with sources shown
How it runs
Define
The job the AI does, and what a good and a bad answer look like.
Prototype
The AI feature tested on your examples before the app is built round it.
Build
App and backend developed with limits, logging and fallbacks.
Evaluate
Accuracy checked against test cases, then released.
Compare the sizes
The standard price for each size. Extras are priced beside it, not hidden behind a call.
| Included | AI prototype€3402 to 3 weeks | AI app MVP€8306 to 10 weeks | Full AI app€1,75010 to 14 weeks |
|---|---|---|---|
| Feature definition and test cases | Included | Included | Included |
| Working prototype app with one AI feature | Included | Included | Included |
| Backend proxy for API calls | Included | Included | Included |
| Evaluation on up to 30 examples | Included | Included | Included |
| Up to twelve screens | Not included | Included | Included |
| User accounts and usage limits | Not included | Included | Included |
| Answers from your own documents with sources shown | Not included | Included | Included |
| Store submission | Not included | Included | Included |
| Up to three AI features | Not included | Not included | Included |
| Admin dashboard with usage and cost monitoring | Not included | Not included | Included |
| Feedback buttons that log weak answers for review | Not included | Not included | Included |
| 30 days of post-launch support | Not included | Not included | Included |
| Order | Select | Select | Select |
Add what you need
Choose the size
Add what you need
Your total€830
ContinueThe standard price, plus whatever you add. Confirmed in writing before anything starts; nothing is charged until you accept it.
Questions
Who pays for the AI usage?
You do, directly to the model provider, on an account in your name. We set limits and show estimated cost per user so there are no surprises.
What happens when the AI gets something wrong?
It will sometimes. We design for that: showing sources, letting users flag bad answers, and handing over to a person or a standard search where confidence is low.
Is user data sent to the AI provider?
Only what the feature needs. We choose provider settings that exclude your data from model training where available and document what is sent. Your privacy policy should reflect it.
Related searches
Related work
One-timeCross-platform mobile appOne codebase for iOS and Android, built in React Native or Flutter and published to both app stores.from €560one-time
One-timeRules-based chatbotA guided chatbot that answers common questions and routes enquiries, with no surprises in what it says.from €290one-time
One-timeMobile app designiOS and Android app screens designed in Figma, with flows, states and a component library ready for build.from €420one-time
Order it
Standardised enough to order on its own. The price is confirmed in writing before anything starts.