Landing Page
Dashboard
Map
Footage

Neutralis

About

Neutralis is a real-time video understanding platform that rethinks how surveillance systems infer identity and escalate incidents. Instead of relying on appearance-based signals or opaque heuristics, Neutralis prioritizes contextual behavior—such as motion, interaction, and event thresholds—to support more accountable and reviewable public-sector workflows.

Built around the idea that surveillance systems don’t just observe people but assign identities with real-world consequences, Neutralis deliberately reduces unnecessary identity leakage. Visual features that do not improve incident detection accuracy are abstracted away, shifting system attention toward behavior that is more relevant to public safety and easier to audit.

Neutralis processes live CCTV streams by segmenting them into short temporal windows and analyzing them through a real-time backend powered by FastAPI. When an event threshold is crossed, clips are sent to the TwelveLabs video understanding API to generate semantic interpretations and structured incident summaries. These outputs are stored as event-centric records—not personal profiles—in a Neon PostgreSQL database, ensuring minimal data retention and clear audit trails.

All inference and review workflows are secured through Clerk-protected API routes, enforcing authenticated access and scoped permissions from ingestion to escalation. By storing incidents instead of identities and preserving footage only when necessary, Neutralis transforms automated perception into something bounded, reviewable, and accountable, rather than opaque and appearance-driven.

Neutralis does not aim to expand surveillance. It aims to make existing surveillance systems more transparent, responsible, and aligned with the realities of public-sector deployment.