Our Features
Six things CounselGuard does for a law firm.
Each one solves a specific job a GC, risk partner, or KM lead already owns — not a generic “AI strategy.” Scroll through, or jump straight to the feature you came for.
AI Tool List
A live inventory of every AI tool your people actually use.
ChatGPT, Copilot, Harvey, in-house tools, the side-loaded extension someone installed last Thursday. CounselGuard catalogs each one with vendor, hosting region, data sensitivity, who at the firm uses it, and which matters it has touched.
- Recognized against our global AI tools catalog — even apps your registry hasn't seen still get caught
- Tagged with vendor, jurisdiction of hosting, and data-sensitivity tier
- Per-user adoption view: who uses what, when, and on which matter
- Set Approved / Restricted / Prohibited per tool — sessions land in the right bucket automatically
Rules by Office
Your jurisdictional obligation map, kept current.
ABA Opinion 512 in US offices, the EU AI Act in Frankfurt and Paris, FLSC guidance in Toronto, SRA expectations in London. Each office is mapped to the rules that actually bind it, and each rule is linked to the firm policies and evidence that demonstrate compliance.
- ABA 512, EU AI Act, FLSC, and SRA frameworks built in
- Office-by-office view of which obligations apply where
- Each obligation linked to the policy, training, and evidence that satisfies it
- Updates surface as a diff when a regulator changes guidance
Jurisdictions covered
Policies and Training Records
Versioned firm policy with approvals and per-lawyer training completion.
Draft a policy, route it through reviewers, approve it, archive it — every transition recorded. Each lawyer's training completion is tracked against the version they were trained on, so when a partner asks 'were they trained,' you have the date and the version.
- Draft → Pending Approval → Approved → Archived, with full version history
- Multi-reviewer approval rounds with reassignment and self-review gating
- Per-lawyer training records tied to the policy version reviewed
- Audit log captures every approval, request-for-changes, and revocation
Activity Capture
A real record of how your lawyers use AI — not survey data.
A desktop agent and a Chrome extension capture actual sessions: which tool, which prompt, which response, on which matter. Sessions finalize through a Gemini sanitize-and-analyze pass that surfaces topic, risk level, and any policy violation.
- Mac and Windows desktop agent plus Manifest V3 Chrome extension
- Transcript reconstruction works across any AI tool — no per-vendor integration
- Session-end analysis flags policy violations and assigns a risk level
- Captures file-only and image-only submissions, not just typed prompts
Audit-Ready Reports
One packet that pulls tools, policies, training, and activity together.
When a client procurement team, a cyber insurer, or a regulator asks how the firm governs AI, the answer is a single export: inventory of tools, approved policies and versions, training completion per lawyer, captured activity, and any incidents and their resolutions.
- Generate by client, by matter, by office, or firm-wide
- PDF and structured export — both signed with a content hash
- Includes audit-log evidence of who approved what, when
- Reusable for client RFPs, insurer renewals, and bar inquiries
Investigations
When something flags, you can drill in, gather evidence, and document the decision.
Open a session, read the actual transcript, attach internal notes and external documents, assign severity, and resolve with a written disposition. The investigation record carries through to the audit log and to any report that touches the matter.
- Drill into the full sanitized transcript for any flagged session
- Attach files, links, and internal notes alongside the AI capture
- Severity and status workflow (Open → Reviewing → Resolved)
- Resolution writes through to the audit log and reports automatically
Investigation workflow
Get in Touch
Make your AI governance defensible
We respond with a brief note on whether we are a fit and, if so, suggest a short call.