The outcome you are walking toward
A hardware spec for one construction site that you can defend in a planning meeting: which rugged edge and network layer the site needs, which presence, vision, and environmental sensing the risk profile calls for, which kit tier matches the site, and the offline behavior the stack is designed to hold when connectivity drops — all as a reference configuration, finalized per site survey.
Read this first: what is staged
Scaffold (BUILD) is a staged rollout — IN_MOTION, not a released inventory. Everything below describes the reference configuration the Scaffold kits are designed around. Where a capability is staged rather than generally released, the wording says "designed to," and a per-site survey finalizes every deployment. There are no prices here, no performance numbers, and no vendor names: kit contents are specified by role and rating, not by brand.
The design principles for a temporary site
The playbook's first principle is the one that shapes everything else: the job site is temporary but the organism must be coherent — rapid deploy, rapid teardown, rapid redeploy. The rest follow from it:
- Safety loops are sacred — they must function offline and under abuse.
- Progress is designed to be visible in near real time, not only at the end of the day or week.
- Crew and equipment presence are first-class signals.
- Harsh environment resilience is non-negotiable — dust, vibration, temperature, power instability, intermittent connectivity.
- Standardized, rugged kits over fragile custom builds.
- Subcontractor boundaries are respected while still feeding the site record.
The site stack, layer by layer
Six layers, specified by role. This is the full reference stack; the kit tiers below are how much of it a given site actually takes.
- Layer 1 — Core edge (mandatory, ruggedized). A rugged edge node — fanless or filtered, wide temperature, shock/vibration rated — with an optional secondary edge on larger or high-risk sites; industrial networking (fiber preferred where possible, bonded cellular as primary or backup); high-endurance power (generator integration, large UPS, or hybrid). Local Mesh, local Synapse, local policy, and the critical agent runtime live here, with Scaffold Field and Sentinel at high priority. The L0–L4 layer model is where these components are defined.
- Layer 2 — Access, presence & identity. Site access control (turnstiles, gates, geofenced mobile credentials, BLE/UWB badges), crew check-in/check-out hardware or mobile-first flows with strong presence verification, visitor and delivery logging, and zone presence for high-risk areas, confined spaces, and heavy-equipment zones.
- Layer 3 — Vision & safety sensing. Rugged cameras covering entry/exit, high-risk work zones, crane and lift areas, material storage, and temporary structures and edges; on-edge models designed for PPE detection, fall-risk and edge proximity, unsafe behavior patterns, and structured progress photography; environmental sensors (gas, noise, dust, temperature) where required; wearable panic / man-down devices for high-risk trades.
- Layer 4 — Equipment & asset tracking. Equipment telematics gateways for heavy equipment, generators, and temporary power; RFID / BLE / UWB tags for tools, high-value materials, and critical assets; utilization and idle-time sensing; geofencing of major equipment.
- Layer 5 — Field compute & capture. Rugged tablets and handhelds for supervisors and crew leads, selective wearable or helmet-mounted capture, digital daily report and safety observation stations, and temporary site boards where useful.
- Layer 6 — Connectivity & resilience. Primary plus failover connectivity (fiber / cellular / satellite), local store-and-forward with large durable buffers, pre-configured rapid-deployment networking kits, and physical security for the edge and network gear.
How the physical layer feeds the agents
Every sensor above exists to produce an event with a priority lane and a named agent behind it. The loops the site is designed around:
Physical event Priority Primary agents
Crew member enters high-risk zone without PPE Critical Sentinel, Scaffold Field
Near-miss or unsafe condition observed Critical Sentinel, Scaffold Field, Atlas
Man-down or panic signal Critical Sentinel, Scaffold Field, Atlas
Environmental threshold breach Critical Sentinel, Scaffold Field
Progress photo + structured update High Scaffold Field, Atlas
Material / tool leaving geofence High Atlas, Sentinel
Daily crew presence vs plan variance High Nexus, Scaffold Field, Atlas
Major equipment idle beyond threshold Medium-High Atlas, Scaffold Field, LedgerScaffold Field and Sentinel live inside these loops; safety and progress are designed to be continuous, not retrospective. What those agents are allowed to do with a signal is bounded by the governed agent boundaries — sensing and alerting escalate, they do not self-authorize.
The strictest offline contract in the lineup
Construction sites frequently have poor or intermittent connectivity, so the playbook holds this vertical to a stricter offline contract than most. The EDGE spec's vertical notes say it in one line: harsh environments demand higher physical resilience and store-and-forward for safety observations. The contract the stack is designed to keep:
Must continue offline:
- Access and presence recording
- Critical safety alerting — local sirens, supervisor devices, wearable feedback
- Local progress and observation capture
- Equipment utilization logging
- Full event buffering with integrity
Degrades gracefully:
- Cross-site intelligence
- Heavy model updates
- Non-critical analytics
Forbidden while offline:
- Loss of safety event data
- Silent dropping of presence or access records
- Policy weakening
On reconnect, reconciliation is designed to be deterministic and auditable, with explicit gap detection if any exists — no silent data loss, ever. The full offline behavior contract, including the buffering rules and the seven-step reconciliation protocol, is owned by the edge runtime offline contract guide; the audit trail & evidence model covers how each buffered observation becomes reviewable evidence.
Kit tiers: match the site, not the wish list
Five standardized kits, all designed for repeated deployment across a network of projects and for rapid redeployment:
- Scaffold Core Site Kit — rugged edge + access/presence + essential cameras + basic tablets. For small to medium sites.
- Scaffold High-Risk Kit — Core + dense vision + wearables + environmental + enhanced alerting. For high-risk or regulated work.
- Scaffold Heavy Equipment Kit — Core + deep telematics + utilization + geofencing. For equipment-intensive projects.
- Scaffold Large Project Kit — multiple edges + full vision + zone control + high connectivity resilience. For major projects and multi-zone sites.
- Scaffold Rapid Deploy Kit — pre-configured, flight-case style, fast set-up and tear-down. For short-duration or emergency sites.
Kit selection follows the project and risk profile assessment, never the other way around. Hardware from these tiers is how the MCV.STORE kits are specified for a build deployment.
Deploy, run, tear down, redeploy
The deployment sequence the kits are designed around, end to end:
- Project & risk profile assessment
- Kit selection and connectivity plan
- Physical rapid deployment and mounting
- Mesh join + policy + safety pack download
- Access, presence, and high-risk zone configuration
- Equipment and wearable onboarding
- Agent activation (Scaffold Field + Sentinel + Atlas + Nexus)
- Supervisor and crew lead enablement
- Live operation with daily evidence reviews
- Demobilization & kit reset for the next project — preserving learnings
Step 10 is the principle in action: the site ends, the organism's record does not. Kit reset preserves what the project learned so the next deployment starts smarter.
What to check before you commit
Three checks turn this page into a defensible spec. Walk the site survey questions against the layer list above — which zones are high-risk, what the power and connectivity reality is, how long the site runs. Confirm the offline contract section against your safety obligations, especially the forbidden list. And ask how conformance evidence applies to the device classes in your kit tier before anything joins a production Mesh. The build field record discipline doc covers the record this hardware feeds, and the Scaffold vertical page is where the staged rollout status lives. For the rollout shape around the hardware, the pilot rollout guide is the only sequence we recommend.