Manako has quietly rolled out a vision AI agent that uses Bittensor subnets to detect objects and events in real time. The platform, aimed squarely at enterprise surveillance, promises to cut through video noise and surface only what matters — a move that could reshape how factories, warehouses, and corporate campuses monitor their operations.
What the Agent Does
The agent runs directly on Bittensor, a decentralized machine-learning network built on the Polkadot ecosystem. Instead of sending video streams to a central server, the AI processes footage at the edge, flagging anomalies, people, vehicles, or specific behaviors as they happen. Manako says the setup improves operational efficiency by reducing false alerts and making data from existing cameras more useful.
For now the company isn't naming specific clients or pilot sites. But the technology targets industries where security and operational oversight overlap — logistics yards, manufacturing lines, retail floors. One advantage of using a subnet is that the AI can be updated without swapping hardware, letting the system learn new patterns over time.
Why Bittensor Matters
Bittensor subnets allow multiple participants to train and run models in a distributed way. For an enterprise surveillance system, that means the detection model isn't locked inside a single vendor's cloud. Manako's agent taps into the subnet's collective compute, which the company says lowers latency and spreads the processing load.
The choice also carries a decentralization pitch: no single entity controls the detection logic, so the system can keep running even if one node goes down. For businesses worried about vendor lock-in or single points of failure, that might be a selling point. Manako hasn't published performance benchmarks, but the architecture suggests the agent can handle multiple camera feeds simultaneously without bogging down a central server.
Competition and Context
Manako isn't the first to marry AI with enterprise surveillance. But most existing systems rely on cloud APIs from Amazon, Google, or Microsoft. Tapping into a decentralized subnet is a different approach — one that could appeal to organizations with strict data-residency rules or those that want to avoid monthly per-camera fees.
The launch comes as more companies look to automate security monitoring. A single guard watching dozens of monitors inevitably misses events. Manako's agent aims to catch those misses, flagging only the clips that need human attention. Whether the Bittensor infrastructure can deliver the reliability enterprises expect is still an open question — the network is younger than Amazon's AWS, and its track record at scale is short.
