Security and compliance

LystSign is a global product with an India-first compliance posture. Here is how it is built, described as plainly as we can.

Certifications

LystSign is designed for SOC 2 Type II and ISO/IEC 27001 readiness. We do not hold a SOC 2 report or an ISO 27001 certificate today and will say so until an independent auditor issues one. A data processing addendum aligned with India's Digital Personal Data Protection Act 2023 is available on request.

Where your data lives

Documents, encryption keys, audit trails, backups and logs are stored in AWS Mumbai (ap-south-1).
Disaster-recovery copies are kept in AWS Hyderabad (ap-south-2), encrypted with a separate key.
If you use a custom signing domain, signer traffic passes through the AWS CloudFront edge in transit; nothing is cached there.
Optional AI features run on Amazon Bedrock in India (Mumbai or Hyderabad) and only after an admin turns them on.

Encryption and isolation

TLS in transit; AWS KMS customer-managed keys at rest for the database, documents and backups, with yearly rotation.
Every workspace is isolated by row-level security enforced inside PostgreSQL, not only in application code.
Documents are converted and sealed by a worker with no route to the internet.

Sign-in and access

Argon2id password hashing and breached-password screening.
Passkeys (WebAuthn) for passwordless sign-in, or authenticator-app two-factor with recovery codes, and re-authentication before sensitive actions.
Roles and permissions per workspace, with admin actions recorded in the workspace audit log.

Monitoring

AWS account activity is recorded by a multi-region CloudTrail with log-file validation, kept in an encrypted, access-logged bucket.
The production profile adds GuardDuty threat detection, Security Hub (CIS and AWS best-practice checks), AWS Config and a web application firewall; high-severity findings page the operations team.
Dependencies, code, infrastructure templates and container images are scanned on every change.

AI features

Off by default. A workspace admin turns AI on, chooses each feature (field detection, summaries and risk flags, signer questions, reminder drafts) and accepts a disclosure; every change is recorded in the workspace audit log.
Requests go to Anthropic Claude on Amazon Bedrock through an India-only inference profile, from our servers over a private AWS endpoint. Amazon Bedrock does not use prompts or replies to train models.
Email addresses, phone numbers, PAN, Aadhaar-like and account numbers are masked before text is sent. Prompt and document text are never written to our logs.
Client-side sealed documents are never sent to AI: the server holds only ciphertext for them.
AI output is a suggestion. Senders review and edit fields and reminder text; summaries, risk flags and signer answers say they are not legal advice, and signer answers cite the page they rely on.
Signer questions are not stored and are rate-limited. Each use counts against a monthly AI credit allowance with a hard cap, so there is no surprise bill.

Evidence and integrity

A hash-chained audit trail for every envelope: sent, opened, viewed, signed, declined, voided.
Sealed, tamper-evident PDFs with a certificate of completion.
Designed against the Information Technology Act 2000 and the Indian Evidence Act framing for electronic records; ESIGN/UETA (US) and eIDAS (EU) are considered in the design but not claimed.

Availability

Application servers run in two availability zones; the production profile adds a Multi-AZ database. Point-in-time database recovery and daily backups, which can be copied to a second region.
Alarms on errors, latency, capacity and background jobs notify the operations team.
A public status page shows current health.

Whether an electronic signature is valid for a particular document depends on the document type, the parties and the governing law; some documents need a specific signature method or cannot be signed electronically. Please take advice for your situation. Report a vulnerability to security@lystsign.com.