Shipping data is confidential. Consignees, freight terms, carrier rates, commodity descriptions , all of it is commercial intelligence. Parsela is built with enterprise-grade security from the first line of code, not bolted on after the fact. This page documents how.
Parsela is in pre-GA with active design partners. SOC 2 Type II framework is in progress. The technical controls required for certification are already implemented; the audit itself is scheduled post-GA.
In freight, merging two shipments that should have stayed separate is a compliance incident, not a data-quality bug. Two customers' cargo appearing on one BL. Two bookings merged because they share a container. One shipper's exports appearing under another's manifest. Any of these triggers customs exposure, regulatory disputes, and, in the worst case, cargo released to the wrong consignee.
Parsela's identity resolution is designed to never make this mistake. It links shipments only on strong, unambiguous identifiers, and treats weaker signals like container numbers, vessel-voyage pairs, and email IDs as evidence to record, never as a reason to merge. When it is not sure, it leaves records unlinked for an operator to join, because a missed link is one click to fix and a false link is a compliance incident.
A booking can split into many shipments. Many bookings can merge into one consolidated shipment. But only with explicit operator action and an AMENDED audit event carrying the reason code (carrier re-issue, customer change, consolidation split, consolidation merge, operator correction), the effective date, and the triggering document identifier.
Every amendment is append-only logged. Rolling back an amendment creates another audit event, history is never silently edited. The compliance officer and the customs broker can reconstruct any shipment's identity history end-to-end without trusting the application layer.
On the field level, every shipment field carries an ownership state (Parsela-proposed, human-edited,
or locked once a downstream event such as customs filing has committed it). When a value changes, a
field_edit_log entry records the old value, new value, actor, source (Parsela, TMS
operator, system), and timestamp. When Parsela and a TMS operator disagree, the conflict surfaces in
a resolution queue, Parsela does not silently overwrite a human edit.
Parsela uses JWT-based authentication with short-lived tokens and server-side revocation. Role-based access control (RBAC) enforces a four-tier authority matrix across every privileged endpoint: Analyst (1), Manager (2), Director (3), Admin (4). Every mutation writes a canonical audit log entry capturing actor identity, target, before/after state, and timestamp.
Parsela processes your shipping documents to extract structured data. Here is exactly what happens to that data at each stage.
Documents arrive via IMAP email polling, web upload, or REST API. At ingestion, Parsela stores the original PDF in customer-controlled object storage (AWS S3, Google Cloud Storage, or on-prem MinIO depending on deployment). Document bytes never transit uncontrolled infrastructure.
Extraction uses frontier AI models under vendor agreements that forbid training on submitted data. For cloud deployments, API calls are made through the provider's standard commercial privacy terms. For customers with stricter data residency requirements, Parsela can be deployed with the hyperscaler's managed AI service (AWS or GCP), keeping document data within the customer's VPC. On AWS VPC deployments, embeddings use Amazon Bedrock in the customer's account; GCP VPC deployments are available on request. See the sub-processors section below for the specific vendors.
Extracted structured data is stored in PostgreSQL (RDS for AWS, Cloud SQL for GCP, or self-hosted) with encryption at rest enabled by default. Original PDFs remain in object storage with server-side encryption. Access is limited by vendor scoping, the multi-tenant architecture ensures operators only see documents from vendors they are explicitly assigned to.
All external traffic uses TLS 1.2+. Internal service-to-service calls within the VPC are also TLS. Webhook delivery to your TMS includes HMAC-SHA256 signatures so your TMS can verify the payload originated from Parsela.
Shipping documents regularly contain personal data (consignee contact names, signatures on PODs). When the PII_REDACTION_ENABLED setting is on, prompts carry pattern-masked identifiers: bank and card numbers, tax IDs, Social Security numbers, email addresses and phone numbers. The setting is off by default, and masking does not detect names or addresses. Document files are not redacted: a PDF or image goes to the model provider unredacted. Prompts and documents are processed under the provider's standard commercial API terms. Anthropic's published policy says "we automatically delete inputs and outputs on our backend within 30 days of receipt or generation", with exceptions that include enforcing its Usage Policy and legal requirements, and its Commercial Terms say "Anthropic may not train models on Customer Content from Services." (Quoted as published by Anthropic, November 2026.) Redaction of personal data based on Microsoft Presidio is planned.
Different customers have different data sovereignty requirements. Parsela ships in four deployment configurations, controlled by environment variable, no code changes required to switch.
Parsela uses a small set of sub-processors. Each is vetted for security posture and has a DPA in place. The list below applies to Cloud (Standard) deployments. VPC deployments use the hyperscaler's AI services in the customer's own account. On-premises and air-gapped deployments, available on request, run with fewer or no external sub-processors.
Parsela's port and place reference data comes from published, versioned sources, each recorded with its edition: UN/LOCODE (United Nations Economic Commission for Europe), the World Port Index (Pub. 150, U.S. National Geospatial-Intelligence Agency), and GeoNames (geonames.org), used under the Creative Commons Attribution 4.0 licence. Country names come from the Unicode CLDR.
If you believe you've found a security vulnerability in Parsela, please email [email protected] with a description and reproduction steps. We commit to acknowledging all reports within 48 hours and publishing a fix timeline within 7 days. Responsible disclosure is appreciated and we're happy to discuss bounty arrangements for substantive findings.
We're happy to walk through the architecture, DPA, and compliance roadmap on a call. Design partners get a more detailed security review as part of onboarding.
Book a security discussion →