Fix the right thing first.
Evidence behind every answer.
PlantGuard turns your plant's own records — inventory, monitoring, backups, lifecycle dates, Active Directory, firewall rules, door access — into prioritised findings with the evidence attached, one Plant Health Score and a plan you can budget. All on one server that never leaves the plant network. It is the security logbook and the advisor that reads it.
- Every finding cites its evidence
- Deterministic rules; local AI only explains
- No cloud, no agents on your PLCs
Designed for isolated industrial environments
- On-premises
- Air-gap ready
- Local AI (optional)
- Evidence-driven
- No mandatory cloud
- Signed offline updates
Each of these is verified against the shipped code and explained, with its limits, in the Trust Center. Explore Security Architecture →
Why PlantGuard
Your tools tell you what happened. PlantGuard tells you what it adds up to.
Your NMS says the switch is up. Your backup console says a job failed. Your asset list says the model is past end of support. Only PlantGuard puts those three facts on the same asset, scores what they mean together, and writes down what to do first — with every fact quoted.
Findings with evidence
Thirty-six deterministic rules read the inventory, monitoring, backups, lifecycle, compliance, firewall, door-access and security-posture data you already collect. Each finding shows the fact, the interpretation, the recommendation and a transparent 0–100 Priority Score you can open with "Why 83?".
One score, six sub-scores
The Plant Health Score rolls open findings into availability, recovery, cyber security, lifecycle, compliance and configuration control. Every point is explained; a dimension without data says not assessed instead of guessing.
Plan, budget, report
Correlated risks name the assets to plan around. The Lifecycle & Budget planner turns lifecycle facts into a year-by-year plan with your own cost figures. Twelve reports export as branded PDF, CSV or Excel — offline and audited.
How it works
Five steps, all on one server inside the plant
The language model, if you enable it, takes part only in the fourth.
- 1CollectGather evidence from the systems you already run — devices, directory, backups, monitoring, endpoint consoles, lifecycle data.
- 2CorrelateConnect the facts across assets, identity, backup, monitoring, lifecycle and security so they land on the same asset.
- 3PrioritizeDeterministic rules decide which findings need attention and score them 0–100, transparently.
- 4ExplainTranslate technical evidence into operational impact — the optional local AI phrases only what the evidence says.
- 5ProveShow the records behind every finding, score and plan, and export them as audited reports.
Correlated risks
The core switch nobody wants to think about
End of support passed, warranty expired, last configuration backup 41 days old, SNMPv2c with a well-known community, five downstream assets. Five facts, five places — already in PlantGuard. Correlated risks turn them into one finding with an ordered plan: backup first, close the exposure, restore the baseline, then replace.
- Evidence quoted from every child finding
- Closes automatically when the children close
- Carried into the Lifecycle planner and the Executive Summary
- Vendor supportended 14 months ago
- Warrantyexpired
- Configuration backup41 days old
- SNMPv2c, default community
- Downstream assets5
Critical infrastructure lifecycle and recovery risk — Priority Score 100.
AI Advisor
Ask it. It answers from the evidence — or says it cannot.
What should I work on first? Which devices still use SNMPv2c? Is 10.0.2.1 healthy? The question is routed to the findings, scores, plans and reports, and the answer cites the finding ids it rests on. The optional local model only phrases that evidence; wording that cites anything PlantGuard does not know is thrown away. When there is no evidence, the answer is exactly: PlantGuard does not currently have enough evidence to determine this.
How the AI is bounded“Start with CORE-SW-01: it carries the correlated lifecycle and recovery risk [PG-2026-000027] and is past vendor support [PG-2026-000001]. HIST-SRV-01 is offline with failing backups [PG-2026-000028]. Assign owners first.”
Illustrative example. Every bracketed id opens the finding and its evidence.
Still the plant's security logbook
One database, one login, one audit trail
The advisor sits on the same records as the twelve modules PlantGuard is built from.
Complete visibility
Switches, servers, HMIs, firewalls and stand-alone assets — site, building and room, vendor, serial, EOL, warranty, owner, criticality, every network card — with Asset 360 and QR labels.
Continuous monitoring
SNMP v2c/v3 and WMI polling, syslog and trap reception, LLDP/CDP topology discovery, scheduled configuration backups and rogue-device scans.
Lifecycle you can budget
Support contracts, hardware end of life and warranty — each entered once — feed the planner, the spare-parts matrix and the Executive Dashboard.
Provable compliance
Group Policy scored against a 34-check baseline cited to NCA ECC-2:2024 and OTCC-1:2022, framework coverage views and waivers with who, when and why.
Framework mappings and assessment capabilities are intended to assist organizations with evidence collection and internal assessment. They do not constitute certification, accreditation, or endorsement by the organizations responsible for these frameworks. PlantGuard.ai does not hold, and does not claim, any certification against IEC 62443, NIST SP 800-82, ISO/IEC 27001, NCA ECC or NCA OTCC.
Integrations
Reads the systems your plant already runs
Every connector syncs live data and matches it to the inventory by IP, MAC or host name — including a device's extra network cards. Unmatched records are reported, never invented into assets.
Endpoint protection
Backup
Identity, scanning & SIEM
The scanner connector is read-only; SIEM forwarding is CEF or JSON over syslog.
Physical access
Network & hosts
Firewall rule import
Read-only, over SSH.
Security built in
Controls you would ask of any OT system
Details and limits for each are in the Trust Center.
Server-enforced permissions
Administrators and read-only operators; each operator sees only the tabs their account allows — and the server refuses the rest.
Two-factor authentication
Offline authenticator codes for local and Active Directory accounts, required per role by policy, with single-use recovery codes.
Confirmation for sensitive changes
Creating, changing or deleting users and changing sign-in policy ask for the administrator's password again.
Encrypted credential vault
Device credentials encrypted at rest, never returned by the API, click-to-reveal with an audit entry for every reveal.
A complete audit trail
Every sign-in, change, export and question recorded on one clock, and forwarded to your SIEM if you want it there.
Signed offline updates
Update packages carry Ed25519-signed manifests and SHA-256 hashes; backups keep their keys only in a password-encrypted bundle.
Early access
Be first in line when PlantGuard.ai ships — Q1 2027
We are finishing validation with pilot plants. Leave your work e-mail and we will send the launch date, the security pack, and an invitation to an evaluation slot. No newsletter, no third parties.
- Launch notice and pricing before public release
- Priority for pilot and evaluation licenses
- Security & architecture pack on request