Security

BatesFlow handles privileged legal documents. We built our architecture from day one around a single principle: the platform operator structurally cannot access your case data. This isn’t a policy — it’s enforced by the system’s design.

What BatesFlow’s AI Does — and Doesn’t Do

BatesFlow does not generate legal arguments, citations, advice, motions, pleadings, or strategy. AI is used only for bounded document operations: OCR, document classification, production-mapping suggestions, demand parsing, and gap surfacing. Every AI output is reviewable and overridable by the lawyer before anything is produced. AI never files, serves, or finalizes anything on its own.

Zero Data Retention

Your documents are never retained by third-party AI providers. Content sent for classification and OCR is processed in real time and immediately discarded — it is not stored, logged, or used for model training. BatesFlow maintains enterprise-tier data processing agreements with all AI providers to guarantee this contractually.

Data Residency

All BatesFlow data — documents, database records, backups, and production outputs — is stored and processed exclusively in the United States.

How Your Data Moves Through BatesFlow

1. Upload — files enter over an encrypted connection.
2. Encrypted storage — stored in AWS S3 in the United States, encrypted at rest with AES-256.
3. AI processing — text excerpts and page images are sent for OCR and classification (see AI scope and AI processing for how this data is handled).
4. Tenant-scoped results — results are written into your firm’s isolated environment; no other firm can see them.
5. Lawyer review & override — your team reviews and can override every classification and mapping.
6. Bates-stamped production — you generate the court-ready production package.
7. Export or delete — you can export your data or request deletion at any time, under your firm’s control.

Three-Layer Privilege Boundary

BatesFlow enforces strict data isolation through three independent layers:

Layer 1 — Tenant Isolation

Every law firm operates in a completely isolated environment. Each firm’s data — cases, documents, productions, user accounts — is scoped to their tenant. There are no shared tables, no cross-firm queries, no accidental data leakage. One firm cannot see, access, or even detect the existence of another firm’s data.

Layer 2 — Role-Based Access Control

Within each firm, access is controlled by role. Firm administrators manage their own users. Attorneys see only the cases assigned to them. Staff members have limited permissions. Every action is logged with the user, timestamp, and IP address.

Layer 3 — Support Access

BatesFlow operators have no standing access to your documents or productions. Support access is off by default. If support is needed, a firm administrator can grant a time-limited, read-only support session. Even then, support access is limited to case metadata — case names, document counts, classifications, and status. It cannot open or read document contents or OCR text, cannot export files, and cannot edit, delete, or run a production. A grant is bound to a single operator, is logged, expires automatically, and can be revoked by the firm at any time.

Encryption

• In Transit: All data is encrypted using TLS 1.3 between your browser and our servers.
• At Rest: All documents and database records are encrypted using AES-256 encryption.
• File Storage: Uploaded documents are stored in encrypted object storage with server-side encryption and access logging enabled.

Infrastructure

• Cloud Hosting: BatesFlow runs on isolated cloud compute with managed database infrastructure.
• Document Storage: Encrypted object storage with access policies and audit logging enabled.
• Background Processing: Document processing jobs run in isolated worker processes.

Malware Protection

Files uploaded to BatesFlow are automatically scanned for malware.

Supply-Chain & Monitoring

BatesFlow uses Socket to continuously monitor its open-source dependencies for supply-chain attacks. Socket inspects every dependency — and every dependency update — for malicious behavior such as hidden install scripts, unexpected network or filesystem access, and obfuscated code, flagging risky packages before they reach production. This protects against compromised or typosquatted libraries being pulled into the application, and security-critical dependencies are kept patched.

Application health and errors are monitored in real time with Sentry, so problems are detected and addressed quickly.

AI Processing

BatesFlow uses enterprise-grade AI for document classification and OCR. Important details:

• Document content sent to the AI provider is not used for model training.
• AI processing uses enterprise-tier APIs with data processing agreements in place.
• Classification results are stored only within your tenant’s isolated environment.
• You can review and override any AI classification before production.

Subprocessors

BatesFlow relies on a small set of vetted subprocessors. Each is listed below with what it touches and where it runs.

Access Controls

• Password hashing using bcrypt with per-user salts.
• Session-based authentication with per-request validation.
• CSRF protection on all state-changing operations.
• Account suspension takes effect immediately — active sessions are invalidated.
• Firm administrators can deactivate users instantly.

Audit Logging

• Comprehensive Logging: Every user action — logins, document uploads, classifications, production runs — is logged with user, timestamp, and IP address.
• Session Management: 30-minute session timeout with per-request validation. No stale sessions.
• Retention: Audit logs are retained for a minimum of 12 months.

Security FAQ for IT

Do you train on our data? No. Document content is processed under a commercial agreement that prohibits using your data to train AI models.

Can BatesFlow employees read our documents? Not by default. Operators have no standing access. A firm admin can grant time-limited, read-only support access limited to case metadata — it cannot open documents or OCR text.

Where is our data stored? In the United States.

Is our data encrypted? Yes — encrypted in transit and with AES-256 at rest.

Do you support export and deletion? Yes — self-serve data export (ZIP) and firm-initiated deletion with a 30-day window.

Do you use AI to give legal advice? No. AI is limited to document operations.

Can a human override the AI? Always. Every classification and mapping is reviewable and overridable before production.

Do you scan uploads for malware? Yes (see Malware Protection).

Incident Response

In the event of a confirmed data breach, BatesFlow will notify affected firms within 72 hours, including details of what data was affected, what steps we’re taking, and what actions (if any) the firm should take.

Compliance

• CPLR Compliance: Production outputs are formatted to meet New York Civil Practice Law and Rules requirements for discovery responses.
• Data Retention: You control how long your data is retained. Deletion requests are processed within 30 days.

Responsible Disclosure

If you discover a security vulnerability, please report it to security@batesflow.com. We take all reports seriously and will respond within 48 hours.

Questions

For security-related questions, contact security@batesflow.com.