— Service
AI integration
When adding an AI feature is on the table but its placement and user value remain unclear, this is the area we address. We make use of provider APIs such as OpenAI and Anthropic, or Apple's own Vision and Whisper frameworks — whichever fits the problem.
— What is included
- OpenAI, Anthropic and Mistral API integrations
- Apple Vision (image-to-text, object recognition)
- Whisper (speech-to-text) and TTS (text-to-speech)
- Vector-database search — converting user notes into queries
- Cost control and response caching
— Process
01
Sanity check
The first question is whether AI genuinely adds value here, or whether it is being introduced for its own sake. If the latter, we deliver conventional software instead.
02
Prototype
A working version of a single feature is produced within 1–2 weeks and tested internally before exposure to users.
03
Optimisation
Response time, cost and error rate are tightened prior to production deployment. A response longer than three seconds typically loses the user.
04
Launch
The capability is delivered as a feature within an existing application or released as a standalone product.
— Frequently asked questions
Who covers the API costs?+
We cover costs during development. At launch they are transferred to you, by which point cost optimisation is in place so that monthly budgets can be reliably estimated.
Is user data sent to OpenAI?+
Only what is strictly necessary. Identifiers such as names and phone numbers are stripped before transmission, or omitted entirely when not required. We treat KVKK obligations seriously.
Let's discuss AI integration
Messages are responded to within the same business day, by the following morning at the latest.