An adversarial test up front.
A boring, well-run program after it.
The assessment is the wedge: it shows the buyer how the site gets robbed, in minutes and dollars. The program is what keeps it from happening, delivered through partners so the company stays small.
Two ways in, depending on who's buying
An EPC that buys an adversarial test of its own site is buying a report that says its security failed. So the full red team goes to owners and lenders; EPCs get a non-adversarial walk-down they can buy the same week.
Physical red-team assessment
owners · IPPs · lenders$7.5k / $14k / $25k · creditable within 60 days
- Walk-down plus a night-time breach attempt under a pre-negotiated rules-of-engagement template signed by every party with site control.
- Law enforcement notified. No-touch rule for energized equipment.
- Pentest-format report, 0–100 risk score, dollarized expected annual loss, layered control plan.
- Patrol and sheriff response times mapped and stated as a limit.
Theft-exposure walk-down
EPC preconstruction$3.5k · creditable · no intrusion
- Laydown-yard layout, spool staging, fence and gate condition, response mapping.
- An auto-generated zone plan from the Site Assessor, ready to price into the bid.
- Sold to estimating and preconstruction so security is in the general conditions, not bolted on.
Do they finish before anyone arrives?
A cable theft is a race between the crew's work time and your detection-plus-response time. Change the layers and response time to see who wins. Timings are illustrative assumptions used by the prototype, not measured results.
Assumed work rate about 90 ft of cable per crew-member per 10 minutes once inside; marking and signage reduce the chance a crew picks the site (shown as deterrence) but do not slow a crew that's already inside.
What we buy, what we build
The company owns no hardware and rolls no trucks. It buys commodity layers and builds only what is differentiated: the evidence and the assessment engine.
Partner categories describe candidate suppliers. LVT, for example, publishes a partner API and integrates with Immix-class platforms (API Evangelist; LVT monitoring). No supply agreement exists with any vendor.
Two linked tools, running on sample data
Both are working front-end prototypes. They show the buyer the pentest wedge and the operational product side by side.
Site Assessor
Draw your perimeter on satellite imagery (or load the sample 150 MW Nevada site), enter site facts, and get a 0–100 theft-risk score, pentest-style attack paths, expected annual loss and a costed, layered control plan with auto-placed camera coverage and honest gaps.
Open Site Assessor →SOC console
A simulated live feed across six sample sites. Triage scores each event, auto-dismisses wildlife, wind and debris with a reason, escalates intrusions with a video-verification card and a dispatch timeline, writes a hash-chained evidence log, and generates an incident report and a monthly attestation PDF.
Open SOC console →Where this doesn't work, and what we do about it
Remote response
Many desert sites are 45–90+ minutes from a deputy. We map it before quoting, write it into the report as a limit and lean on deterrence where response exceeds 30 minutes.
Coverage, not a dome
The 4-zone default covers laydown yards, spools and active MV and substation areas. Full-perimeter coverage of a 1,000-acre site is a separately priced upsell, and the Assessor draws the gaps.
Scope and liability
Detection and verification only, with a liability cap. The central station is primary custodian of video. Incident support beyond 10 hours is billed hourly.
Licensing
Nevada's NRS 648 likely requires a license for the assessment itself. The plan budgets its own license through a qualified manager; counsel confirms before spend.
AI claims
A high auto-dismiss rate for false positives is a target to measure in pilots, not a fact. The prototype's triage is a scripted simulation.
Drones
Detection only. Private drone mitigation is generally unlawful, and we won't sell it.