AI in the product
AI development inside software people already use
AI features in the workflow — capture, retrieval, a suggestion the user can undo — not a research lab.
We add AI where it changes a job: a reminder from “dentist Tuesday 3pm,” search in your own words, a focus timer that suggests 20 minutes because that is how you work. BrainSync AI is the production example. We do not train custom foundation models. If the data to support a feature does not exist, we will say so.
If you need the app around the feature, that is custom app development, mobile, or a web application. If the AI is one job in a first version, that is MVP development.
Delivery is remote from the Auckland studio. How we work from Auckland.
What you get
A feature in the workflow
Capture, retrieval, or a suggestion the user can undo — not a separate science project.
An explanation when we recommend something
Ghost Mode says why a session length was picked. Black-box scores are out.
Integration on stacks we already ship
Kotlin and cloud on BrainSync; TypeScript/Node if the feature lives in a web app.
A cut if it does not earn the first screen
We will not pad an MVP with a chatbot nobody asked for.
AI in shipped work
BrainSync AI is production: capture, reminders, search. InspectAI is a 2-week MVP with AI/ML listed — a smaller slice, not the same product.
FAQ
Can AI be part of an MVP, and how long does it take?
Yes, if it is one job. InspectAI and WaitlistAI are published as 2-week MVPs with AI/ML in the stack. We do not quote model-training programmes.
Will you take this if we need a custom large language model?
We do not train custom foundation models. We build AI features into products using the stacks in our case studies.
Want AI in the product?
Describe the sentence a user would say, and what should happen next. We will say if it is worth building.