Our long-term research and product direction. Dates are aspirational targets, not commitments.
PurenetX's long-term vision is to bring context-aware intelligence into camera systems, enabling privacy-preserving, on-device decisions that help prevent harmful visual content before it is created.
Here's our path to make it real:
The technologies described on this page represent our long-term vision and research roadmap. They are NOT all available in today's prototype. We are transparent about what we've built (see Current Status), what we're building (Phase 2), and what we're researching (Phase 3). Nothing on this page is a promise. These are research directions.
What we've validated so far. The concept is proven — the real inference engine comes next, built with partners.
What we're actively building. Timelines are targets, not guarantees.
Production-grade SDK for enterprise platform integration, refined through design partner pilots. Supports embedding into camera pipelines, video processing systems, and security platforms.
The core ML model, trained on real design partner data. This is the heart of Phase 2 — 8–12 weeks of co-development per partner, beginning once pre-seed funding closes.
Native implementations enabling integration directly into mobile camera apps and consumer platforms. Required for the long-term OEM vision.
Model compression and optimization for lower-power edge devices — IoT cameras, embedded systems, entry-level smartphones.
12-week co-development partnerships with 2–3 enterprise design partners. Partners shape the model with their data, get SDK framework access, co-design sessions, and research credit. Pilots with the real model: Q2–Q3 2027.
Long-term research vision. These represent where the technology needs to go — not what it does today. No OEM partnerships currently exist.
Integrate the Context Intelligence Layer into live camera workflows, enabling privacy-preserving safety decisions before capture occurs. Requires camera API partnerships with device manufacturers.
Integration into iOS, Android, and Samsung camera apps as a default safety layer. Requires deep OEM and platform partnerships that do not yet exist.
Customizable decision policies allowing platform operators to define their own safety thresholds and responses: Allow, Warn, Blur, Block, Escalate.
Integration into device cameras at the hardware level. Over two billion devices ship with cameras annually — the long-term TAM, not the near-term go-to-market.
None of these capabilities exist in today's prototype. Each represents a distinct research and partnership challenge requiring close collaboration with device manufacturers.
Connecting the Context Intelligence Layer to real-time camera preview streams without impacting device performance.
Model compression for real-time inference on constrained mobile and embedded hardware without cloud dependency.
Hardware Abstraction Layer integration — requires direct collaboration with device manufacturers and OS vendors.
Working within Apple and Google's camera API constraints to embed decision logic at the capture layer.
Mechanisms to act on contextual safety decisions before image capture completes — platform-level access required.
Direct integration with Samsung, Apple, Google camera systems requires formal OEM partnership agreements.