Skip to main content
For OEM cybersecurity leadership

OEM CSMS without manual evidence reconstruction

One connected CSMS for ISO/SAE 21434, UNECE R155, GB 44495 and EU CRA — covering type approval evidence, federated supplier execution, and CVE-to-V‑SOC response in hours, not weeks.

30-min scoping call No slides — bring your architecture Mutual NDA available; not required First win in ≤6 weeks

One ECU family, one OEM template, the rest of your stack untouched.

In plain terms: ThreatZ gives your CSMS team one traceable model across suppliers, programs, vulnerabilities, controls, tests, and type-approval evidence — without replacing your existing EA, ARXML, DOORS, Jira, or supplier portal.

Smallest pilot scope: one architecture, one supplier slice, or one CVE response scenario.

Three regulators. One model. Audit-ready evidence on demand.
REGULATORS ONE CSMS MODEL EVIDENCE OUTPUTS UNECE R155 Type approval · EU/UK/JP GB 44495 China · enforced Jan 2026 EU CRA Vuln handling Sep 2026 · CA Dec 2027 ThreatZ CSMS knowledge graph 8 WORKFLOWS TARA SBOM CONTROL TEST EVIDENCE RISK DESIGN Type approval pack R155 + GB 44495 + EU CRA Supplier evidence Federated · scoped sub-tenants CVE → V‑SOC disclosure R155 §7.2.2 · internal triage in hours Click any node, the assessor sees the trace.

One model, three regulators, every supplier — authored once, audit-ready in every jurisdiction.

In pilots and evaluations with cybersecurity teams at

BMW Vector Informatik Foxconn Brose Preh Neusoft Reach
Operational outcomes

What changes when the chain is connected

Four operational outcomes OEM cybersecurity teams report in first-cycle programs. Outcomes vary by SBOM coverage, supplier mix, and existing toolchain.

14 d → <4 hrInternal exposure assessment + draft disclosure packageCoordinated supplier disclosure and OTA campaign decisions follow normal timelines. Outcomes vary by SBOM coverage.
~80%EA → system-node auto-mappingDrift detection within ~1 hour
8 / 1Workflows on one platformCounted against typical 5-vendor reference stack (TARA + GRC + ticketing + SBOM + portal).
3Regulators, one modelR155 + GB 44495 + EU CRA covered by one connected model
The OEM CSMS reality

Three regions. Five evidence formats. Dozens of suppliers. No single traceable chain.

TARA in documents. Risk in spreadsheets. SBOMs in CycloneDX, SPDX or custom exports. Controls in a wiki. V-SOC findings disconnected from design-time risk. When the assessor asks for one variant’s cybersecurity case, the team reconstructs truth from tools never built to behave like one CSMS.

Three regions, three timelines

UNECE R155 type approval, GB 44495 — China type-approval enforced from January 2026 (phased rollout for in-production types through 2027), EU CRA vulnerability handling and incident reporting obligations from 11 September 2026; full conformity assessment from 11 December 2027. Same evidence, three regulator formats, three audit cadences.

Five evidence formats, no chain

TARA in Word, risk register in Excel, SBOMs from forty suppliers in CycloneDX exports, residual-risk acceptance in a SharePoint folder, controls in a wiki page nobody updates after the audit closes.

Two weeks per public CVE

A CVE drops Tuesday. UNECE R155 §7.2.2 vulnerability handling and Annex 5 Part B threat mitigation demand a disclosure assessment. By Friday: 200-supplier coordination via Teamcenter, Windchill, vendor portals — none of which carry the cybersecurity model. Manual SBOM joins, attack paths reconstructed from memory. Two weeks later, you have a partial answer for one region.

Where the cost appears

The risk is not missing documents. The risk is not being able to prove the chain when it matters.

A new CVE drops. The team needs to know which components are affected, which ECUs and variants include them, which scenarios and controls are relevant, whether tests prove mitigation, and whether a disclosure package is owed in each region. Without a connected CSMS model, that answer becomes emails, manual SBOM joins and late-night evidence packaging.

The same CVE, two operating models. Outcomes vary by SBOM coverage and supplier mix.
Step Current setup — manual reconstruction ThreatZ — one connected model
CVE landsWatch the feed, manually triage which programs care.Ingested into the graph; PURL / CPE / SHA-256 match runs automatically.
Supplier exposure40–200 supplier coordination via Teamcenter, Windchill, vendor portals — none of which carry the cybersecurity model. Reconcile PDFs by hand.Supplier-delivered CycloneDX / SPDX already bound to components.
ECU & variant bindingArchitecture knowledge lives in two senior engineers’ heads.Component → ECU → variant → release, queryable.
R155 §7.2.2 disclosure~14 days, late-night packaging, partial answer for one region.R155 & GB 44495 disclosure drafts in under 4 hours (internal exposure assessment; coordinated supplier disclosure follows normal timelines).
One graph, eight workflows

ThreatZ connects the work of CSMS — not just the documents of CSMS.

Six platform positions, each paired with what delivers it in the product.

Eight workflows · one knowledge graph
THREATZ Knowledge Graph CSMS · 8 workflows Design Architecture Governance Policy · RBAC TARA Risk · CAL SBOM CycloneDX · SPDX Testing PenTest · SAST Compliance R155 · GB · CRA Collaboration Suppliers · CIA Operations V‑SOC · Annex 5

Eight workflows, one CSMS knowledge graph. Every node, every edge, every claim resolves to the same model — no integration tax between Design, Governance, TARA, SBOM, Testing, Compliance, Collaboration, and Operations.

Connected, not fragmented

Eight workflows — Design, Governance, TARA, SBOM, Testing, Compliance, Collaboration, Operations — on one knowledge graph. One tenant, one model, one identity surface. Versus the typical OEM stack of five disconnected vendors plus the integration tax that exceeds license cost.

Architecture-centric

Import your system model once. ARXML (network plane) and MATLAB System Composer + Simulink (function + behavior plane) ingest as orthogonal sources joined at the asset graph. TARA, SBOM, test coverage and compliance reporting all derive from the same model. Supported: AUTOSAR Classic 4.2.x / 4.3.x / 4.4.x System Description + ECU Extract; AUTOSAR Adaptive SOME/IP service descriptions; DBC fallback for legacy CAN. Bus types: CAN, CAN-FD, FlexRay, Ethernet/SOME-IP, LIN.

Federated supplier execution

Your suppliers don't just receive your templates. They receive a workspace inside your tenant, scoped by RBAC at org / project / entity. Your methodology, your review cadence, your evidence schema — live status, not quarterly PDFs.

CVE response under four hours

First-cycle programs report ~14 days collapsing to under 4 hours for internal exposure assessment + draft disclosure package: PURL / CPE / SHA-256 match against SBOM, component-to-ECU-to-variant binding, risk re-score, and R155 §7.2.2 + Annex 5 Part B + GB 44495 disclosure packages drafted for cybersecurity sign-off. Coordinated supplier disclosure and OTA campaign decisions follow normal timelines.

Sovereignty by deployment

Private cloud or air-gapped on-premise. Customer-owned data plane. China-resident deployment for GB 44495. Air-gapped option for defense-loaded programs. SAML / OIDC / LDAP / AD for identity. Audit-log export to your SIEM. See /security/.

AI that traces, doesn't generate

The AI Recommender doesn't generate a TARA opinion. It traces relationships in your knowledge graph — CVE to component to ECU to variant to threat scenario to control to test. AI accelerates. Humans approve. Every claim has a graph link the auditor can follow. AI Recommender runs in customer tenant; no telemetry or training data egress in private-cloud / on-prem / air-gapped deployments. Model keyed to project knowledge graph, not a generic LLM.

Bring your current evidence flow to a 30-min map

No slides — we walk your architecture and show where the chain breaks for the assessor.

Book a CSMS Traceability Review

“The architecture-mapping piece — auto-suggesting bindings from our EA model — saved weeks per program. ThreatZ replaced four spreadsheets and a Jira board with one queryable graph the assessor could walk.”

Head of Vehicle Cybersecurity
European OEM · multi-program portfolio

Practitioner quoted under NDA. Named references available after the scoping call.

The traceable chain

From global CVE to vehicle impact, risk decision, and V-SOC evidence.

A new vulnerability drops. Walk the chain. Every node is an actual module; every edge is an actual graph relationship.

CVE → V‑SOC chain
01 CVE Ingest 02 SBOM Component 03 Supplier CIA / DIA 04 ECU Binding 05 Variant Re‑score 06 Threat Scenario 07 Control §10 08 Test Proof 09 V‑SOC Incident 10 Disclose R155 · GB

Ten queryable nodes, one connected chain. Feeds ingested: NVD, CISA KEV, EUVD (ENISA), Auto-ISAC advisories, vendor PSIRT, customer-configured. VEX statements consumed alongside SBOM. From CVE ingestion through SBOM match and binding to disclosure packaging — every edge is a graph relationship the assessor can walk.

01

Ingest

Ingest from NVD, CISA KEV (Known Exploited Vulnerabilities), EUVD (ENISA EU Vulnerability Database), Auto-ISAC advisories, vendor PSIRT feeds, and customer-configured feeds. VEX (Vulnerability Exploitability eXchange) statements consumed alongside SBOM. Disclosure flow aligned with ISO/IEC 29147 (coordinated vulnerability disclosure) and ISO/IEC 30111 (vulnerability handling processes), customer’s PSIRT cadence preserved.

02

SBOM match

PURL, CPE, SHA-256 / MD5 match across supplier CycloneDX + SPDX.

03

Binding

Component → ECU → variant → release, via Architecture Mapping Studio. EA → ThreatZ mapping via XMI 2.4 export + MDG technology, with auto-suggested system-node bindings (brownfield projects start at 30–40% match and improve as schema profile establishes; ~80% on clean greenfield models).

04

Re-score

Feasibility (5 factors) × Impact (S/F/O/P) → CAL re-check.

05

Surface

Affected attack paths and controls surfaced from the graph; AI accelerates, humans approve.

06

Record

Decision typed into the graph — role, identity, timestamp, Policy Manager rules.

07

Draft disclosure packages

R155 §7.2.2 + Annex 5 Part B + GB 44495 disclosure packs auto-drafted for cybersecurity sign-off — from the same graph that fed ISO 21434 evidence.

Federated supplier execution

Your suppliers execute your process. On your platform. With scoped access.

One tenant, scoped sub-workspaces, one auditable graph — no more quarterly PDF reconciliation.

How it works

A Tier-1 is invited via the Supplier Manager persona. RBAC scopes them to one project (or one entity inside it). Your methodology, TARA templates, Risk Treatment policy, and Security Catalog of controls are pre-loaded.

What they deliver

CycloneDX or SPDX SBOM ingested directly. Threat scenarios in your STRIDE + automotive taxonomy. Attack paths scored against the 5 ISO 21434 feasibility factors you mandated. Tests linked to your requirements and controls.

What you see

Live status across every supplier and every program. The R155 type approval evidence packet auto-assembles from supplier-delivered evidence as it lands — with the lineage the assessor needs.

Federation pattern — OEM & suppliers on one platform
OEM TENANT One knowledge graph RBAC: org · project · entity SUPPLIER A Your TARA template SBOM · CIA · tests SUPPLIER B Your methodology Live status, not PDFs SUPPLIER C Your review cadence CycloneDX / SPDX SUPPLIER D Auto-evidence For R155 type approval

Suppliers don't email PDFs — they receive a workspace inside your tenant, RBAC-scoped to their entity. Your methodology, your templates, your review cadence. Live status. Audit evidence auto-assembles in one graph.

ISO/SAE 21434 evidence chain

From asset to incident. Every clause linked.

The chain the assessor walks. Every node is a real module in the product, not a diagram label.

ISO 21434 evidence chain — every clause linked
Asset Damage Scenario Threat Scenario Attack Path Feasibility CAL 1–4 Impact Risk Goal §9 Requirement §10 Control §10 Claim Test Verification §10 Validation §11 Work Product §6.4.8 Field Monitoring §13 + R155 Annex 5 Incident loop

Click any node, the assessor sees the trace. One highlighted path shown here; the model carries every node.

Asset → Risk

Asset, Damage, Threat (STRIDE + automotive), Attack Path, Feasibility (5 ISO 21434 factors), CAL 1–4, Impact (S/F/O/P), Risk.

Goal → Test

Goal (§9 Concept), Requirement (§10 Product Development), Control (§10, from Security Catalog), Claim, Test (PenTest, VulnScan, SAST, Config Review, Functional Security), Verification (§10.4.2), Validation (§11 Cybersecurity Validation).

Work product → loop

Cybersecurity Case (§6.4.8, auto-generated as a §15 distributed-activity work product when suppliers are in scope), Field Monitoring (§13 Operations & Maintenance + R155 Annex 5), Incident, looping back into Risk. Every arrow is a queryable graph edge — not a diagram label.

If you already built one

Keep the portal. Replace the disconnected model behind it.

Layered architecture — portal stays, model swaps
EA ARXML MATLAB DOORS Codebeamer Jira SBOM scanners CycloneDX / SPDX V‑SOC / SIEM YOUR EXISTING PORTALS & DASHBOARDS OEM internal portal SharePoint GRC Custom dashboards FEDERATION LAYER REST GraphQL SAML OIDC LDAP Git Sync SIEM export THREATZ CSMS KNOWLEDGE GRAPH Asset Threat Control / Test design TARA SBOM testing compliance ops

Your team’s user-facing portals stay. ThreatZ becomes the cybersecurity knowledge graph behind them, federated via REST / GraphQL / SAML / OIDC / LDAP / Git Sync.

APIs first, UI optional

REST and GraphQL APIs expose every entity in the graph. SAML / OIDC SSO federation against your IdP. Audit-log export to your SIEM. Git Sync for documentation versioning. RBAC at org / project / entity.

On-premise or air-gapped

Customer-owned data plane. China-resident deployment for GB 44495. Air-gapped option for defense-loaded programs. Your portal stays the user-facing surface; ThreatZ owns the model behind it.

Sovereignty & deployment

Private cloud, on-premise, or air-gapped for sovereignty-sensitive programs.

ThreatZ supports customer-owned data planes, China-resident deployment for GB 44495, private cloud for enterprise programs, and air-gapped installation for high-sovereignty environments.

Deployment options — what stays inside your control under each mode.
Capability Private Cloud China-Resident On-Premise Air-Gapped
IdentitySAML / OIDC against your IdPSAML / OIDC, in-country IdPSAML / OIDC / LDAP / ADLocal LDAP / AD
Audit logsExport to your SIEMIn-country SIEMYour SIEM, your retentionOn-box, exportable
Data plane ownershipCustomer tenantChina-resident, customer tenantCustomer infrastructureCustomer infrastructure, isolated
Supplier accessFederated sub-workspacesFederated, in-countryFederated via VPN / DMZFederated via offline package exchange
Compliance package generationR155, ISO 21434, EU CRAGB 44495, R155 projectionAll regionsAll regions, offline rendering

For OEM procurement teams with AWS commitments: ThreatZ Team and Pro tiers are available via AWS Marketplace as a procurement channel (EDP burndown eligible). Same ThreatZ, two deployment modes: AWS Marketplace for authoring and pilots, private cloud or on-premise wherever vehicle data lives. Enterprise programs remain direct.

UNECE R156 (Software Update Management System / SUMS) is the matched twin VTA to R155. ThreatZ supports R156 evidence — release-baseline binding, update-impact analysis, and OTA campaign audit trail — alongside R155 evidence.

China-resident deployment supports MLPS 2.0 (等保2.0) network security graded protection, PIPL (Personal Information Protection Law) data-export controls, and the Automotive Data Provisions (2021). In-country compute + in-country IdP + CATARC-compatible compliance reporting.

Air-gapped federation via signed CycloneDX + VEX bundle exchange + tagged JSON evidence packages; manual import scheduled by site cybersecurity officer.

Capability maturity: what ships in every deployment vs what's scoped per customer.
Available todayScoped per customer
TARA workflows + risk-relationship graphEnterprise Architect integration (XMI + MDG)
SBOM ingestion (CycloneDX, SPDX)MATLAB / System Composer model mapping
R155 + ISO/SAE 21434 + GB 44495 compliance reportingDOORS / Codebeamer / Polarion bidirectional sync
Federated supplier execution + RBACJira / ALM workflow sync
Risk + evidence linkingV-SOC / SIEM integration

Sovereignty, supplier mix, identity provider — we’ll scope it

One call, one architecture. We map the supplier set against private-cloud, on-premise or air-gapped before any commitment.

Scope a 6–8 week pilot
Audit-grade

Six questions your R155 assessor will ask

Six questions an external assessor (KBA / VCA / RDW / UTAC / CATARC / TÜV SÜD / DEKRA) will ask — current setup answer vs. ThreatZ answer. Audit questions sourced from R155 §7.2 process audit, ISO/SAE 21434 §6 cybersecurity management, ISO/SAE PAS 5112 audit framework, and observed VDA cybersecurity audits.
Assessor question Current setup answer ThreatZ answer
Cybersecurity case for variant X, build Y, date Z (§6.4.8) Senior engineer reconstructs from Git history, SharePoint, and three TARA spreadsheets. 5–10 days. Variant + release baseline scoping: weakness trees unique per project/variant/release. Time-travel query against the graph. Minutes.
Walk me through one §8–§11 chain end-to-end, including the §15 distributed activities The chain exists across four tools and one shared drive. Reconstruction = email thread + screenshots. Asset → Damage → Threat → Attack Path → Feasibility → CAL → Risk → Goal → Requirement → Control → Claim → Test → Verification. Click any node.
How is each supplier's CIA (Cybersecurity Interface Agreement) / DIA (Data Interface Agreement) rolled into the OEM case? Suppliers email signed PDFs. OEM cybersecurity engineer manually maps to the OEM case. Quarterly. Supplier executes in your tenant. CIA / DIA artefacts live as graph objects, bound to components, version-controlled, signed.
Residual-risk acceptance record: who, what role, when? Email approval, sometimes a PDF signature, often missing for risks accepted three years ago. Policy Manager defines who can accept which risk class. Acceptance is a typed event in the graph: role, identity, justification, timestamp. Immutable.
R155 Annex 5 monitoring evidence, past 12 months, with dispositions Pulled together from ticketing, V-SOC notes, and CVE response emails. Coverage gaps are common. Operations workflow links incidents to component, severity, MTTR (Mean Time To Resolution), disposition; feeds back into the risk model.
Project the cybersecurity case into the GB 44495 package format for China homologation Six weeks of consulting to reformat into the Chinese regulatory schema. Whole-team task. One model, many projections. GB 44495 templates ship with the platform; the package renders from the same graph that fed your R155 file. China-resident deployment supported.
Honest scope

Focused, not fragmented

Six categories ThreatZ deliberately doesn't live in. We integrate with the tools you already trust; we don't try to replace them.

Not a replacement for Enterprise Architect

ThreatZ imports software architecture from EA. UML modelling stays where it is. Architecture Mapping Studio is the bridge.

Not an SBOM scanner

ThreatZ consumes CycloneDX and SPDX. The vulns-scanner service matches components to feeds; generation stays with your scanners.

Not exploit generation

Defensive cybersecurity engineering data, not red-team automation. No PoC pipelines, no offensive tooling.

Not a CI/CD pipeline

Integrates via Git Sync, Codebeamer, DOORS — doesn’t run your build. ALM and pipeline tools stay where they are.

Not a mandatory user-facing portal

Can run behind your internal portal as the model layer. APIs first; UI optional.

Not multi-tenant for your vehicle data

Private cloud or on-premise — air-gapped supported. Customer-owned data plane. Vehicle data stays inside your tenant.

Practitioner voices

What OEM cybersecurity leadership tells us

“We stopped reconstructing evidence the week before each audit. The cybersecurity case for any variant is a query now, not a project.”

Head of CSMS Programs
European OEM · multi-region type approval

“ThreatZ eliminated the duplication and gave us confidence both documentation sets were consistent and complete. We achieved European type approval months ahead of schedule.”

VP Cybersecurity
Chinese EV manufacturer · vehicle program

Practitioners quoted under NDA. Named references available after the scoping call. Outcomes vary by program scope, supplier mix, and existing toolchain.

See ThreatZ in action

The graph is real. The trace is queryable.

Eight workflows, one connected model. Click an asset, see every threat, control, test, and disclosure pack downstream. Bring your architecture and we walk one trace together in 30 minutes.

ThreatZ platform dashboard showing the CSMS knowledge graph: assets, threats, controls, tests, and evidence linked structurally.

Project status, supplier sub‑tenants, CVE chain, evidence packs — one tenant, one queryable model. Book a Traceability Review →

Bring one program. We will trace the evidence flow end-to-end.

In 30 minutes, Uraeus maps your current CSMS evidence flow across one architecture, one supplier slice, or one CVE response scenario. We identify where the trace breaks, which evidence packages can be generated from a graph, and what a federated tenant would look like for your supplier set.

Best attendees: Head of Vehicle Cybersecurity + one TARA lead + one type-approval owner. 30 minutes.

30-min, no slides Mutual NDA available Open formats: ReqIF, ARXML, CycloneDX, SPDX, CSV