Honest about today
HerdGuard's MVP is cloud-assisted. We do not currently claim built-in WAN/LTE, off-grid operation, or on-device (edge) AI as present-tense features.
Technology
A phased path from cloud-assisted MVP to a HerdGuard-trained platform.
HerdGuard is currently in early development. We start with a focused MVP, validate with real properties, build a verified dataset, and earn each capability before claiming it.
Principles
What we will and will not claim today.
Every feature on this site is either part of the current MVP or clearly identified as roadmap. We avoid language that implies guaranteed protection, off-grid operation, or built-in cellular hardware.
Still images, not video
Detection is performed on still images captured by internet-connected farm cameras. Real-time video pipelines are out of scope for the MVP.
Dataset-first thinking
Every reviewed event is a unit of long-term value. The verified dataset is what supports a HerdGuard-trained model on the roadmap.
Provider-neutral architecture
The vision provider is an implementation detail. The platform is designed so a HerdGuard model can run side-by-side and gradually take over on supported categories.
Technology & roadmap
A phased path from cloud-assisted MVP to a HerdGuard-trained platform.
HerdGuard is currently in early development. Initial deployments are designed for internet-connected camera locations. Future roadmap items may include hybrid inference, edge AI options, additional connectivity kits, and proprietary model development.
- Phase 1
Cloud-assisted MVP
Still images from internet-connected farm cameras are analyzed by an external inference provider. HerdGuard handles event creation, alert rules, and history.
- Phase 2
Human validation
Owners and operators review and correct events. Confirmed outcomes form a structured, verified dataset for each deployment.
- Phase 3
HerdGuard model side-by-side
A HerdGuard-trained model runs alongside the third-party vision provider for evaluation and gradual hand-off on supported categories.
- Phase 4
Hybrid inference & edge AI options
On-device inference for supported camera classes is evaluated as a roadmap option where it meaningfully reduces latency, cost, or bandwidth. Not part of the current MVP.
- Phase 5
Commercial platform & API
Multi-site management, partner integrations, and an API for farms, ranches, and rural-tech partners.
Inference is currently performed by an external vision provider. The provider name will be disclosed when a finalized commercial relationship permits.