Clients now treat their law firms the way they treat any vendor holding sensitive data. Corporate clients send security questionnaires and outside counsel guidelines. Bar rules and ABA opinions describe what reasonable care for client information looks like. Software sits in the middle: the portal, the intake system, the document tool and the AI assistant are where the questions are answered or failed. We build legal software with those controls designed in and documented.
This page is for IT directors, CISOs, general counsel and partners who own client security commitments. It is about the software we build and run, not firm wide IT. For related work see legal AI development, legal software integrations and the software development for law firms overview.
The ethics guidance behind the requirements
The American Bar Association’s Model Rules and formal opinions shape how firms think about client data, and many states adopt similar rules. Below is a plain summary for planning software. It is not legal advice, the wording is ours, and you should read the primary texts and your own state’s rules before relying on any of it.
| Source | What it addresses, in short | What software can contribute |
|---|---|---|
| Model Rule 1.6(c) | Reasonable efforts to prevent inadvertent or unauthorized disclosure of, or access to, client information | Access control, encryption, logging |
| Rule 1.6, comment 18 | Reasonableness is judged by factors such as sensitivity, likelihood of disclosure, and the cost and difficulty of safeguards | A documented risk based choice of controls |
| Formal Opinion 477R (2017) | Securing communication of client information; the standard is fact specific, and very sensitive matters may call for extra precautions | Encrypted channels, secure portals instead of email attachments |
| Formal Opinion 483 (2018) | Lawyers’ duties when a data breach happens, including responding and informing current clients in the situations the opinion describes | Logs and monitoring that show what happened and whose data was affected |
| Formal Opinion 498 (2021) | Virtual practice and the reasonable safeguards remote work requires | Secure remote access, managed sessions |
| Formal Opinion 512 (2024) | Generative AI: competence, confidentiality, communication, supervision and fees | Data boundaries and review steps for AI tools |
None of these lists a required technology. They ask lawyers to understand the risk and act reasonably. Our job is to give the firm software where the reasonable choice is easy to make and easy to show.
Outside counsel guidelines and security questionnaires
Corporate clients turn their own security standards into contract terms for the firms they hire. Typical questions: Is data encrypted in transit and at rest? Is multi factor authentication enforced? Who can access our files, and how is that reviewed? Are access and changes logged? Which vendors touch our data? How fast will you tell us about an incident? Where is our data stored?
A firm cannot answer these well if its custom software was built without them in mind. We start projects by collecting the questionnaires and guidelines the firm actually receives, then treat each question as a requirement with an owner and a test.
Identity: SSO, MFA and offboarding
We connect the applications we build for a firm to the firm’s identity provider, such as Microsoft Entra ID or Okta, over SAML or OpenID Connect for login. That puts multi factor authentication, password policy and account lifecycle in one place. When someone leaves, disabling them in the directory removes access everywhere. External users like clients and co-counsel get separate accounts with their own MFA and a narrower scope.
Access control, matters and ethical walls
Role based access is the floor. Legal work needs access by client and matter on top of it, plus exceptions: ethical walls that screen a lawyer from one matter, temporary access for a contract attorney, read only access for an auditor. We model these as data, enforce them on the server, and apply them everywhere the data can surface. That includes search, exports, notifications and AI assistants, which are the places where walls often leak.
Where isolation between clients matters most, separate storage helps. In a multi-tenant rebuild we gave each account its own PostgreSQL schema, so a query cannot reach another tenant’s data even if application code is wrong; see the multi-tenant case study.
Audit logs
Audit logs answer the questions that come after something goes wrong: who opened this file, who exported that list, who changed the permission. We log authentication events, record access, changes, exports, permission changes and administrative actions, with user, time, source and target. Logs go to storage that application users cannot alter and are kept for a period your policy sets. If Opinion 483 situations ever arise, these logs are where the facts come from.
Encryption, keys and data residency
Data is encrypted in transit with current TLS and at rest in databases, file storage and backups. Keys live in your cloud account’s key management service, not in our hands or in code. Some clients require data to stay in a specific country or region; we pin storage, backups and processing, including any AI processing, to those regions and document it. Infrastructure detail is on our AWS, Azure and DevOps pages.
For AI tools, residency and confidentiality often mean keeping models inside the firm’s environment. Our legal research RAG assistant ran entirely inside the client AWS VPC with a private model endpoint, and no document content was sent to an external API. The RAG case study describes the architecture.
Privilege, retention and legal hold
Privileged material needs care beyond ordinary confidentiality. We tag documents and records so privilege status follows them into search, exports and AI context, and so a privileged item is not shared through a portal by mistake. Retention is policy driven: closed matter data is archived or deleted on the schedule the firm sets, and a legal hold flag stops deletion for records under hold until it is released. The schedules come from your firm and its counsel; the software enforces them and records what it did.
Vendors, subprocessors and backups
Client questionnaires rarely stop at your own systems. They ask which vendors can see client data. A modern legal application may rely on a cloud host, an email service, an e-signature tool, an error tracker and an AI model provider. We keep that list current for every system we build, note what data each one receives, and configure tools such as error trackers so they do not capture document contents or form fields. Backups get the same encryption and region rules as live data, and we restore them on a schedule, because a backup nobody has restored is only a hope.
SOC 2 and ISO 27001
Clients often ask whether a firm or its vendors hold SOC 2 or ISO 27001. SOC 2 is an attestation report produced by an independent auditor and is common with US clients. ISO 27001 is an international standard for an information security management system, with formal certification. Which one a firm pursues depends on its clients.
We build software to support those programs: controls that map to common framework requirements, evidence that is easy to export, and documentation your auditors can read.
How a security engagement runs
Most projects start with an existing system and a pile of client requirements. We review the architecture, map data flows, list gaps against the questionnaires and guidelines you receive, and rank fixes by risk. Then we build the controls into the product and write the documentation. For new systems the same work happens before the first sprint, which costs far less. If the work includes a new public website, the same access and logging approach applies to its forms and admin; see law firm website design. For broader architecture reviews, see our software architecture service.