Security
Last updated 6 August 2026
You are handing us commercial agreements — cap tables, deal terms, employment arrangements, sometimes your own clients’ confidential material. This page describes what we actually do about that, and what we do not yet do. We would rather you find the gaps here than in a questionnaire six months from now.
Isolation between customers
Separation is enforced in the database itself, not only in application code. Every row carrying customer data is subject to row-level security policies keyed to the owning tenant, so a query issued in one customer’s session cannot return another customer’s rows even if application logic were wrong. This is the control we consider most important, because it is the one that does not depend on every future code change being correct.
What an advocate can see
A panel advocate can see a matter only once it has been assigned to them. Unassigned advocates do not see your matter, your documents, or your identity. Available work is offered to the panel without exposing the parties behind it, so an advocate decides whether to take a matter before they learn whose it is.
Once assigned, the advocate sees what they need to review the document, and every action they take on the matter is recorded.
Audit trail
Actions on a matter — creation, intake changes, assignment, review, approval, download — are written to an append-only audit log with the actor and timestamp. If it ever matters who did what to a document and when, that record exists.
Encryption
- All traffic to and from the platform is encrypted in transit over TLS.
- Data at rest is encrypted by our infrastructure provider (Google Cloud) using their managed encryption.
Payment security
We never see your card details. Payment is taken on Razorpay’s hosted checkout, which is PCI DSS certified; card number, expiry and CVV are entered on their systems and are never transmitted to or stored on ours. We hold only a transaction reference and the amount. We do not store card numbers in any form, tokenised or otherwise.
AI processing
Drafting and clause review use a third-party AI provider. Two things we commit to:
- Your documents are not used to train third-party AI models.
- We do not sell, license or publish your documents, and we do not use one customer’s document as source material for another customer’s document.
Where processing happens
Our application and database currently run on Google Cloud in the United States (us-east1), and document text is sent to our AI provider, also in the United States. We are stating this plainly because it is the sort of thing that should not be discovered later. Payment data is the exception and stays with Razorpay in India.
| Provider | Purpose | Location |
|---|---|---|
| Google Cloud Platform | Application hosting and database | United States (us-east1) |
| Anthropic | AI drafting and clause review | United States |
| Vercel | Web application hosting and delivery | United States |
| Razorpay Software Private Limited | Payment processing. Card details are entered on Razorpay's hosted checkout and never reach our servers. | India |
| Resend | Transactional email delivery | United States |
| Cal.com | Scheduling. Holds the name, email and meeting time of anyone who books a call. | United States |
| Google (Sign in with Google) | Authentication, where you choose to sign in with a Google account | United States |
Access and retention
Access to production systems is restricted to named engineers and is logged. Documents and intake data are retained for 7 years and erased on request within 30 days, subject to the advocate record-keeping exception described in our Privacy Policy.
What we do not have yet
We are an early-stage company and we would rather be accurate than impressive:
- We do not currently hold SOC 2 or ISO 27001 certification. If you need one to proceed, tell us — it affects how we prioritise.
- We have not yet commissioned an independent third-party penetration test.
- Our infrastructure runs in the United States. Moving primary processing to India is on our roadmap but has not happened, and we will not claim it until it has.
Reporting a vulnerability
If you believe you have found a security issue, email security@lextronai.com with the details. Please give us a reasonable opportunity to fix it before disclosing it publicly. We will acknowledge within 48 hours, and we will not pursue action against anyone who reports a genuine issue in good faith and does not access or alter other customers’ data.