ClearPoint Logic

Trust at ClearPoint Logic.

Audit-grade evidence. Vendor-neutral governance. Independent system of record.

Request our security package
ComplianceAI BOMEvidence & auditData residencyDocumentationFAQ
01Compliance posture

Frameworks we map to.

We treat compliance as a deliverable, not a marketing claim. Below is the current state of each framework, with realistic timelines. Detailed mappings and audit reports are available under NDA in our security package.

SOC 2 Type II

Controls mapped
Controls mapped

Security, availability, and confidentiality controls mapped to the Trust Services Criteria. Independent attestation comes later; we won’t imply a report exists before it does.

Controls crosswalk in placeDetail

HIPAA

Controls mapped
Available on Enterprise

Administrative, physical, and technical safeguards mapped to HIPAA Security Rule. Business Associate Agreement available.

BAA on Enterprise contractsDetail

HITRUST CSF

In progress
Mapping in progress

r2 assessment scoping underway. Control mapping to v11.3 covering NIST CSF, ISO 27001, and HIPAA overlays.

Validated assessment H2 2027Detail

ISO 27001

Planned
Planned 2027

ISMS scope defined. Stage 1 readiness audit planned. Annex A control alignment in place.

Stage 1 audit plannedDetail

NIST AI RMF

Controls mapped
Govern · Map · Measure · Manage

Platform controls mapped to all four NIST AI RMF functions. Crosswalk document available under NDA in the security package.

AI RMF 1.0 crosswalkDetail

EU AI Act

Supported
High-risk system support

Logging, transparency, human oversight, and risk management features support customers deploying high-risk AI systems under Title III.

Article 12, 13, 14, 15 controlsDetail
02The AI Bill of Materials

What ships, signed.

An AI Bill of Materials (AI BOM) is a signed manifest of every component that makes up an agent: the model and version, the prompts, the tools and their permissions, the datasets, the libraries, the operator identity, and the policies in force at run time.

Without an AI BOM, you cannot answer the simplest audit questions: what model produced this decision, on which prompt, with what tool access, for which user? Every CPL agent ships with one. Every change is a new signed version.

Vendor-neutral
Same format whether the agent runs on Microsoft, Google, Anthropic, or BYOA.
Cryptographically signed
Tamper-evident. Verifiable offline by auditors and regulators.
Diff-able
Two AI BOMs can be diffed. Promotion is a reviewable change.
Exportable
JSON, CycloneDX-compatible. Your evidence system, your retention.
aibom.compliance-sentinel.3.4.1.json
Signature verified
{
  "aibom_version": "1.2",
  "agent": {
    "id": "agt_compliance_sentinel",
    "name": "Compliance Sentinel",
    "version": "3.4.1",
    "package_digest": "sha256:9f2e1c…7a3b"
  },
  "passport": {
    "id": "pp_8a2c91…",
    "issuer": "anchor.cpl",
    "issued_at": "2026-03-12T14:08:21Z"
  },
  "models": [
    {
      "vendor": "anthropic",
      "name": "claude-sonnet-4.5",
      "role": "primary_reasoning",
      "fingerprint": "claude-sonnet-4.5-20260218"
    }
  ],
  "prompts": [
    { "id": "pr_intake_v8",   "digest": "sha256:c1…" },
    { "id": "pr_summarize_v3","digest": "sha256:b2…" }
  ],
  "tools": [
    { "id": "salesforce_read",   "scopes": ["account.read"] },
    { "id": "evidence_writer",   "scopes": ["bucket.append"] }
  ],
  "datasets": [
    { "id": "kb_policies_2026q1", "rows": 1284, "pii": "redacted" }
  ],
  "dependencies": [
    { "name": "adk-go", "version": "0.42.0" }
  ],
  "policies": ["pol_hipaa_baseline_v6","pol_pii_redact_v2"],
  "signature": {
    "algo": "ed25519",
    "value": "MEUCIQDk…rA=="
  }
}
03Evidence and audit

Replayable runs. Signed traces.

Every agent run produces a signed evidence record: inputs, decision context, tool calls, approvals, model outputs, and the AI BOM in force at the time. Records are append-only, retained per policy, and exportable.

Signed telemetry
Per-step traces signed with tenant-scoped keys. Tamper-evident.
Decision context
The exact prompts, retrieved context, and tool responses that produced an output.
Audit replay
Reconstruct any past run with the same AI BOM. Useful for incident review and discovery.
Retention controls
Per-tenant, per-agent retention. Legal hold supported.
Evidence record · ev_71c8a9e2
Compliance Sentinel · run_3f12 · 2026-03-12
Sealed & verified
14:08:21.044INPUT
POST /v1/agents/agt_compliance_sentinel/run
op:usr_a91 · req:req_3f12
14:08:21.118POLICY
pol_hipaa_baseline_v6 · ALLOW (pii_check ok)
envelope: anchor.policy
14:08:21.402MODEL
claude-sonnet-4.5 · 1842 in / 318 out tokens
fingerprint: 20260218
14:08:21.488TOOL
salesforce_read · account.read · 200 OK
scope verified · 142ms
14:08:21.612APPROVAL
dual_control_required · routed to risk@
awaiting human-in-loop
14:08:48.901APPROVED
usr_b22 · approved · note: “checked exposure”
reversible: yes
14:08:48.953OUTPUT
evidence_writer · bucket.append · 200 OK
sealed
14:08:48.981SIGNED
ed25519 · MEUCIQDk…rA==
record_id: ev_71c8…
04Data residency and BYOK

Your keys. Your region. Your network.

Standard tenancy is multi-tenant on Google Cloud with logical isolation, AES-256 at rest, TLS 1.3 in transit. Enterprise tier adds the controls below.

BYOK

Customer-managed encryption keys via Google Cloud KMS or external HSM (AWS KMS XKS pattern). Per-tenant key hierarchy. Revocation cuts access to data and evidence.

Data residency

Region pinning available for US (us-central1, us-east4), EU (europe-west4), and UK (europe-west2). Evidence and memory stay in region; cross-region replication is opt-in only.

Private connectivity

Private Service Connect, VPC Service Controls, customer VPN/IPSec. Egress allowlist for tool endpoints. No data plane traffic over the public internet on Enterprise.

Tenant isolation

Per-tenant encryption keys, per-tenant pgvector schemas, per-tenant signing keys for evidence and AI BOMs. No shared model fine-tunes across tenants.

PII handling

Redaction at ingest with deterministic tokenization. PII never sent to model providers without an explicit, logged policy decision. Right-to-erasure workflows.

Backup & DR

Daily encrypted backups, point-in-time recovery to 7 days standard, 35 days on Enterprise. Multi-region DR with documented RTO/RPO under NDA.

05Documentation

Documents and resources.

Public documents are linked directly. Documents marked under NDA are bundled in the security package and shared after a brief mutual NDA exchange.

06Frequently asked

Common security questions.

The questions enterprise security teams ask us most often. If yours isn’t here, email [email protected].

Subprocessors process customer data on our behalf. The current list, effective dates, change feed, and API are maintained in the public subprocessor matrix.

We notify customers 30 days before adding or replacing a subprocessor; subscribe from the matrix to track changes without relying on a static marketing-page copy.

Have a security question?

Email our security team directly, or start a conversation through the contact form and we’ll route you to the right person.

Email [email protected]Contact us