Questions, answered

Lexlaw is an AI workspace for teams that work with sensitive documents, such as law firms, clinics and businesses. You can use the leading AI models on contracts, medical records and client files, mask personal data before it reaches the AI provider, and share reusable templates (skills) with your team.

Yes. The Spark plan is free for small teams, with a monthly usage allowance and no credit card. Paid plans add more people, more usage and features like data privacy mode, and each starts with a free trial. The pricing page shows the details.

PDF and Word documents, CSV and other text files, and code. Audio and video recordings are transcribed for you, so the AI can work with what was said.

No. Lexlaw runs in your browser on desktop and mobile, and you can add it to your home screen like an app.

Not if you don’t want it to. Lexlaw’s personal data guardrail swaps sensitive values for placeholders such as [PERSON_1] or [MRN_1] before a message is sent, and puts the real values back in the reply. Administrators choose which of 65 kinds of personal data to mask, and where.

With data privacy mode on, chats can also only use models with zero data retention and no training on your data.

No. Lexlaw doesn’t train AI models. In data privacy mode, chats are limited to models whose providers don’t keep your data or train on it.

Every account needs a second factor: a passkey or an authenticator app, with backup codes. Passwords must be at least 12 characters. People are signed out after a period of inactivity and after a maximum session length, and sessions end immediately when an account is disabled or its role changes. Organizations can also sign in with Google or Microsoft, still with a second factor.

Your chats are yours. Projects can be shared with specific people or teams, with view or edit access. Administrators can open others’ work when they need to, and every time they do it’s recorded in the audit log.

Sign-ins (including failed ones), two-factor and session events, administrative changes, opening someone else’s chats, file downloads and exports. Each entry records who, when and from where, but never the text of messages or documents. The log is append-only, and admins can filter it and export it to CSV.

It goes to a recycle bin, where it can be restored. When the scheduled purge is on, items older than the retention period are deleted for good, including every stored version of uploaded files.

Provider API keys, tool credentials and other secrets are encrypted with AES-256-GCM before they’re stored. Files are uploaded and downloaded through short-lived signed links, with an access check before every download.

HIPAA compliance depends on how a covered entity or business associate uses a service and on the agreements in place, so no software is “HIPAA compliant” on its own. Lexlaw provides data privacy mode, a HIPAA-ready configuration built for health information, designed to be used together with Business Associate Agreements. Talk to us about your setup.

Chats can only use models with zero data retention and no training. People are signed out after at most 15 minutes of inactivity and 12 hours in total. Deleted items are kept for 90 days before purging, the audit log is kept for at least six years, and emailing chats is limited to members of your organization.

Once it’s on for an organization, it can’t be switched off. It’s included on the Max and Enterprise plans.

Yes. The health section covers patient and doctor names, medical record numbers, patient IDs, and NPI and DEA numbers, alongside personal identifiers like dates of birth, Social Security numbers and addresses. Names and some ID numbers use the optional Presidio detector.

Models from the leading providers, grouped into model sets with friendly names such as Fast and Expert. The pricing page lists the sets on each plan. Enterprise organizations can also connect their own providers, including Azure OpenAI and models running on their own servers.

Auto picks a model for each message, based on what the message needs and what each model can do, so quick questions get a fast answer and hard ones get a stronger model. You can always pick a model yourself.

Yes, on Enterprise. Your organization adds its own providers and models, and can build its own model sets. Keys are encrypted at rest.

Yes. Recordings are converted to compact audio in your browser and only the audio goes to the transcription model. The transcript goes through your guardrails before it’s stored, and the AI works with it like any document. Transcripts don’t label speakers yet.

A reusable template: instructions for the AI plus a starter prompt for a job you repeat, such as reviewing a contract or building a medical timeline. Pick a skill, attach your file, and get the same thorough result every time.

Yes. Keep a skill to yourself, share it with your organization, or have administrators publish organization-wide skills. Teams can be limited to the skills they need, and colleagues can use an organization’s skills without seeing the instructions behind them.

Open it, copy its instructions and starter prompt, and paste them into a new skill in Lexlaw. Each skill is also a Markdown file in the same format Lexlaw uses for its built-in skills.

On every plan except Enterprise, each person adds a monthly allowance of credits to a shared pool for the organization. A credit is one cent of AI usage at our model rates. When the pool runs out, chat either pauses until the 1st or switches to the lightest model, depending on the plan.

Every paid plan starts with a free trial; the pricing page shows how long. Spark is free with no time limit.

Your own AI providers and models, MCP tools, your own guardrails, monthly dollar budgets per organization, team or person, and usage shown in provider dollars, plus data privacy mode. Pricing is tailored to your organization.