Matter management · eBilling · spend

Know what outside counsel is costing you — before the invoice arrives.

CounselTrack runs matter intake, invoice review and spend management for corporate legal departments. Every line of every bill is read against your billing guidelines. A lawyer still makes the call.

Register your organization, then invite your team. No card, no sales call.

100%

of invoice lines checked against your guidelines, before a lawyer opens the bill

0

database connections held by the AI layer — it reads through the same API you do

1

database per organization, so one customer's matters cannot be queried with another's

2

approvals on anything touching money or privilege: the model proposes, a person decides

Matter intake

A matter that is classified right the first time.

Guided intake that narrows as you answer. Practice area picks the matter types, so a contract claim cannot be filed as a trademark and the reporting downstream is worth reading.

  • Counterparty, governing law, business unit and cost centre from day one
  • Sequential references your team can quote: MTR-2026-00118
  • Privileged matters restricted to the matter team at the moment they are opened
Invoice review

Every line read against your guidelines. Nothing deducted without a lawyer.

LEDES, PDF and CSV land in one line-item model. Deterministic rules run first — those are reproducible line by line and you can show a firm exactly why. Only then does the model draft the narrative review and the reduction it recommends.

  • Rate cards, staffing rules and billing guidelines per firm
  • Every recommendation traceable to the rule or the narrative behind it
  • Approval thresholds, routing, and straight-through processing for small bills
AI access

Your lawyers can work from the AI client they already use.

CounselTrack publishes an MCP server. Point Claude, ChatGPT or any other client at it and ask in plain language. It is not a second way into your data — it calls the same API the application calls, carrying the identity of the person asking, and it holds no database connection at all. An agent sees exactly what you see and nothing more.

  • Privileged matters stay invisible unless you are on the matter team
  • Every tool call recorded in the audit log and marked as an AI client
  • Writes are propose-then-commit: a person confirms before anything changes
Security

Four rules we do not bend.

One source of truth

The application owns the data. Nothing reaches the database except its own domain services — not the AI layer, not an integration.

Privilege enforced once

A single filter decides what you can see. Screen, API and AI client all go through it, so they cannot drift apart.

AI never decides alone

On money and on privilege a person approves. The model drafts the recommendation; the workflow engine holds the gate.

Your matters train nothing

The provider is swappable and isolated per organization. Customer data is not training data, and the contract says so.

Running today

  • Organization sign-up, invitations and roles
  • Administrator view of people and activity
  • Matter intake, list and detail
  • MongoDB Atlas, one database per organization
  • MCP server with three working tools
  • Automated deploys, five checks gating every push

Being built

  • ·Invoice ingestion: LEDES, PDF and CSV
  • ·Deterministic compliance rules engine
  • ·Budgets, accruals and forecasting
  • ·Single sign-on with Microsoft and Google
  • ·Outside counsel portal

CounselTrack is in preview. This list is the honest state of it, updated as work lands.

Set up your legal department in a few minutes.

Register your organization, invite your team, open your first matter. Your data lives in its own database from the moment you sign up.