On October 20, 2026, Elite, the financial-operations platform that traces its roots to 1947 and processes more than $160 billion in billing transactions across roughly 60,000 monthly users, announced that its 'Elite Intelligence' connectors are live in the Claude marketplace and Microsoft Marketplace. The connectors are built on Anthropic's Model Context Protocol (MCP), the same standard that lets AI assistants query external systems without exporting or replicating data. For a practicing defender, the news matters less for what Elite does and more for what it confirms: querying your system of record from inside Claude or Copilot is no longer experimental. It is table stakes. The question for criminal-defense practitioners is what that looks like when the system of record is a discovery dump, not a billing ledger.
Key points
- Elite, a 79-year-old legal financial-operations platform serving firms billing $160B+ annually, shipped MCP connectors to Claude and Microsoft Copilot on October 20, 2026.
- Model Context Protocol (MCP) is rapidly becoming the default way legal software vendors expose governed, read-only data to general-purpose AI assistants.
- Law firm buyers consistently cite confidentiality and access control as the top barrier to AI adoption, which is why Elite's pitch centers on 'existing permissions remain authoritative.'
- No defense-side evidence review competitor currently ships an MCP integration; the gap is visible and growing.
- Evidence Copilot already provides an org-scoped, read-only, revocable, fully audited MCP server with a no-token OAuth flow at /integrations, letting defenders ask case questions inside Claude Desktop, ChatGPT, Claude Code, or Cursor.
Elite's connectors let authorized users ask questions about clients, matters, invoices, budgets, billing, and work in progress inside Claude or Microsoft Copilot and receive answers grounded in the firm's live financial data. The connection is read-only and respects each user's existing Elite permissions, so no new visibility is granted. CEO Mark Dorman framed the move as extending 'governed AI access' into the tools firms already use, rather than building another standalone AI destination. The strategic point is straightforward. Law firm finance staff no longer need to context-switch into a separate system, run a report, and interpret it to answer routine questions. They can ask in plain English inside the chat tool they already have open. For a category that has spent decades optimizing work-to-cash workflows, this is a meaningful shift in where the answer lives. The governance language is the part to watch. Elite emphasizes that existing permissions remain authoritative, that the connector is read-only, and that no data is replicated. That vocabulary mirrors what skeptical legal buyers have been asking for since the first wave of legal AI tools landed in 2023 and 2024: prove that confidentiality and access control are not being traded away for convenience.
Anthropic introduced MCP in late 2024. Two years later, it has become the default way vertical software vendors expose governed data to general-purpose AI assistants. Research tools, practice management platforms, and now financial-operations systems have all shipped MCP connectors to Claude and Copilot. The pattern is consistent: read-only access, permissions preserved, audit trail intact, no token management required for the end user. For law firms, the buyer concern has never been capability. Modern AI can summarize, search, and draft competently. The concern is governance: who can see what, whether answers are traceable to source data, and whether access can be revoked. Elite's announcement leans directly into that concern, and the framing is now the vocabulary the market expects. When a vendor pitches 'AI for your firm,' the first questions from general counsel and ethics advisors are about permissions, audit, and revocability. Search interest reflects the shift. The keyword 'claude for lawyers' has grown roughly 3,029% year over year to about 480 monthly searches, according to DataForSEO data referenced in industry research. That is the wedge MCP is opening: lawyers want to use the AI tools they already have, pointed at their own governed data.
Here is where the Elite news becomes directly relevant to criminal-defense practitioners. Elite's connectors serve firm finance: billing, matters, budgets. They do not serve the workflow that consumes most of a defender's time before trial, which is evidence review. A typical felony discovery dump might include 40 hours of bodycam footage, multiple jail-call audio files, a 911 recording, interrogation room video, dozens of police reports, photographs, and scanned PDFs with no text layer. The task is not to summarize a budget. The task is to find every inconsistency between what an officer wrote in a report and what the bodycam actually shows, to flag every custodial-interrogation moment that may bear on a Miranda issue, to identify every Brady or Giglio candidate, and to redact protected content before discovery turnover. No defense-side competitor currently ships an MCP integration. JusticeText, Reduct, Vidizmo, CaseGuard, CoCounsel, Sonix, Casefleet, Axon Evidence, and TrialPad all operate as standalone platforms or as tools without governed AI assistant connectors. The free DIY stack (local Whisper plus NotebookLM) breaks at case scale: no diarization quality control, no citations, no redaction, no audit trail. The gap is not theoretical. As MCP becomes standard in legal tech, 'no MCP integration' becomes a visible shortcoming that competitors will have to answer on comparison pages and in procurement reviews.
Evidence Copilot ships an MCP server that lets defenders query their cases from Claude Desktop, ChatGPT, Claude Code, or Cursor. The connector is read-only, org-scoped, revocable, and fully audited. Setup is a standard OAuth flow: paste the connector URL, click Connect, sign in, and allow. There are no tokens to manage. The practical workflow looks like this. A defender is preparing for a preliminary hearing and wants to know every place in the bodycam where an officer's narrative diverges from what the footage shows. Instead of scrubbing through 40 hours of video manually, they ask the AI assistant connected to their Evidence Copilot case: 'Find every moment in the bodycam where the officer's report says the suspect was Mirandized before the search, but the video shows otherwise.' The assistant returns answers with clickable citations that seek the exact second of video or the page of the PDF. Every AI output is marked AI-generated and pending human review. Originals are immutable. The same connector supports the broader evidence-review workflow. A law clerk can ask the assistant to draft a suppression motion outline based on the consent and search-scope moments in the bodycam, with both sides quoted and cited. An investigator can ask the assistant to build an exhibit list tied to real files. A public defender can email a question to their case inbox and receive an answer with citations and deep links. This is the defense-side version of what Elite shipped for firm finance. The difference is the data domain: evidence review, not billing. The shared vocabulary is governance: read-only, permissions preserved, audit trail, human review required.
Elite's governance language offers a useful checklist for evaluating any legal AI connector, whether it serves finance, evidence, or practice management. First, is the connection read-only? If the connector can write to your system of record, the risk profile changes fundamentally. Second, do existing permissions remain authoritative? The connector should not grant new visibility to anyone. Third, is access revocable? If a user leaves the firm or the case, can their connector access be turned off without affecting anyone else? Fourth, is there an audit trail? Every query, every answer, every citation should be logged. Fifth, is AI output clearly marked as AI-generated and pending human review? The product should never make legal conclusions on the user's behalf. These five criteria are not aspirational. They are the floor that Elite's announcement implicitly establishes for any vendor pitching 'governed AI access' to legal buyers. Defense practitioners evaluating evidence review tools should apply the same checklist.
Elite's MCP launch is not a product defenders will use directly. It is a signal that the legal-AI integration standard has matured, and that governance is the vocabulary the market now expects. For criminal-defense practitioners, the practical question is which evidence review tool meets that standard. Evidence Copilot is the alternative on the defense side. The workspace handles transcription with speaker diarization, OCR for scanned documents, AI-suggested review signals on ingest, frame-accurate video redaction, multi-camera sync, and grounded case briefs. Every answer cites the exact source moment. Every AI output is marked for human review. Published pricing starts at $39 per user per month for Solo, $99 per user per month for Firm, and $750 per attorney per year for PD Office, with no feature tiers and a 14-day money-back guarantee. The MCP server is live now at /integrations. Defenders who want to see what governed AI access looks like for evidence review can book a 20-minute walkthrough at evidencecopilot.com/book or start on the Solo plan at evidencecopilot.com/signup.
Elite's October 2026 announcement is a useful proxy for where legal AI is heading: into the chat tools practitioners already use, governed by the permissions they already have, with an audit trail they can hand to a client or a bar investigator. The defense-side version of that shift is already here. The question is whether your evidence review tool meets the same governance floor, or whether you are still exporting spreadsheets and scrubbing through bodycam manually.