Google updates ML Kit to let third-party developers add Google's Smart Reply and other natural language processing features to their Android and iOS apps
Announced at I/O 2018, ML Kit brings Google's machine learning to mobile devices. Available for Android and iOS apps …
Context & Ripple Effects
Smart Reply has been traveling outward from Google's own products for a year: Area 120's Reply experiment carried the feature into other chat apps in early 2018, but only on Android, and the October Firebase update had already deepened ML Kit's on-device vision capabilities with better facial recognition.
This update completes the arc by opening the language side of the kit: any third-party developer on Android or iOS — not just Google's own apps — can now drop Smart Reply and other NLP features in without building a model or running a backend.
First-order effects
- Mobile developers get pre-trained Smart Reply and NLP features as drop-in components on both platforms, removing the build-versus-buy decision for conversational basics that previously required Google's Cloud natural language APIs (in beta since 2016) or custom work.
- The Android-only limitation of the Reply experiment dissolves — iOS apps gain access to the same Google-trained language models natively.
Second-order effects
- Apple's ecosystem becomes a distribution channel for Google's models, putting Google ML inside iOS apps where Apple offers no equivalent turnkey reply/suggestion kit — a quiet competitive wedge.
- Simple NLP workloads shift from per-call cloud APIs toward free on-device inference, pressuring the pricing of Google's own Cloud Natural Language offering for low-volume use cases.
Third-order effects
- If the pattern holds, Google's strategy is to make its trained models ambient infrastructure — embedded in millions of third-party apps via SDKs rather than reached through cloud endpoints — which entrenches Google as the default intelligence layer on rival platforms.
- Developer expectations reset accordingly: on-device ML features become table stakes in mobile SDKs, forcing every platform vendor to ship comparable kits or cede the layer.
The trend: Google is distributing its machine learning as embeddable developer infrastructure, shifting intelligence from cloud API calls to on-device SDKs that run inside competitors' platforms.