The shift is real
A year ago, "AI-native" meant sprinkling a chatbot into your app. Today it means rethinking the entire product surface from the ground up — how data flows, how users interact, and how the UI itself can respond to intelligence.
We've been building in this space since day one with intyx.ai and the dynamic_intyx Flutter package. Here's what we've learned.
Start with the data, not the model
Every AI feature we've shipped started with a clear answer to: what data does the model need to be useful?
For intyx.ai, the answer was structured tabular data — CSV and JSON. Once we had clean ingestion, the AI layer almost wrote itself.
The MCP protocol changes everything for mobile
The Model Context Protocol lets AI agents call tools and read context from your app at runtime. For Flutter, this unlocks something genuinely new: an agent that can render widgets, update state, and guide users — without you hardcoding any of it.
That's the core idea behind dynamic_intyx. The Flutter widget tree becomes a surface the AI can control.
Ship small, learn fast
Both products started as weekend experiments. The version we shipped publicly was iteration 8 or 9. Don't wait for perfect — get real users on it and let the feedback shape the roadmap.
Questions? Reach us at hello@bmnova.com.