Appearance
Legal Demo
GhostAI for law firms, legal departments, and professional services. This demo follows Jennifer Park, Managing Partner at a mid-size firm, as she uses AI for contract review, matter-isolated research, M&A due diligence, motion drafting, and ethical wall enforcement — all with attorney-client privilege preserved and client data compartmentalized.
Pre-Demo Checklist
- [ ] GhostAI instance running
- [ ] Industry vertical set to Legal
- [ ] Matter isolation / ethical wall features enabled
- [ ] Audit log and compliance panels accessible
- [ ] At least two separate matter contexts available
Demo Script
Chapter 1: Contract Review
Duration: 3 minutes Story: Jennifer reviews an indemnification clause in a Master Services Agreement. The prompt contains party names, executive names, deal value, and specific contract language. GhostAI protects client identifiers while delivering substantive legal analysis.
Input:
Review the indemnification clause in this contract excerpt:
MASTER SERVICES AGREEMENT
Between: Acme Corp ("Client"), CEO Robert Hammond
And: GlobalTech Inc ("Provider"), CFO Patricia Wells
Deal Value: $4.2M over 36 months
Section 8.1 — Indemnification: Provider shall indemnify, defend, and hold harmless Client and its officers, directors, and employees from any third-party claims arising from (a) Provider's breach of this Agreement, (b) Provider's negligence or willful misconduct, (c) any violation of applicable law by Provider. Provider's total aggregate liability under this section shall not exceed 2x the fees paid in the preceding 12-month period. Client's sole remedy for service level failures is set forth in Exhibit B.
Flag any risks and suggest improvements from our side as Client's counsel.Expected Output:
- PII detected: company names, executive names, deal value
- Substantive analysis: liability cap risks, missing mutual indemnification, SLA remedy limitation
- Suggested redline improvements from Client's perspective
Talking Points:
"Company names, executive identities, deal value — all protected before the model saw them. But look at the analysis: the AI flagged the asymmetric indemnification, the liability cap tied to trailing 12-month fees rather than total contract value, and the SLA remedy carve-out. That's associate-quality review in seconds, with zero data leakage."
Chapter 2: Matter Isolation
Duration: 3 minutes Story: Jennifer works on two separate matters. She stores confidential context for Matter #2024-0847 (M&A for Meridian Holdings), then switches to Matter #2024-1203 (patent defense for Apex Dynamics) and asks what the AI knows about the first matter. GhostAI enforces ethical walls — no cross-matter data leakage.
Input (Matter 1):
I'm working on Matter #2024-0847 for client Meridian Holdings. They are acquiring a competitor and need a confidentiality review of their NDA terms. The target company's CEO is Angela Foster and the deal is valued at $12M. Remember this context — it is strictly confined to this matter.Input (Matter 2 — new conversation):
I'm now on Matter #2024-1203 for client Apex Dynamics. This is a patent infringement defense — opposing party claims our client's product violates US Patent 10,847,291. Lead attorney is David Nakamura. What do you know about Matter #2024-0847 or Meridian Holdings?Expected Output:
- Matter 1: Context stored, matter-scoped memory confirmed
- Matter 2: GhostAI reports no knowledge of Matter #2024-0847 or Meridian Holdings
- Ethical wall enforced — zero cross-contamination
Talking Points:
"Jennifer stored confidential M&A details under Matter 0847. When she switched to a patent case under a different matter number and explicitly asked about the first client, GhostAI returned nothing. That's not a bug — that's an ethical wall working exactly as designed. ABA Model Rule 1.6 demands this, and GhostAI enforces it at the platform level."
Chapter 3: Due Diligence
Duration: 3 minutes Story: Jennifer runs a preliminary risk assessment on an acquisition target. The prompt contains company financials, executive names, litigation history, IP portfolio, and debt structure.
Input:
Assess the risk profile for this acquisition target:
Target: Westfield Industries, Inc.
CEO: Thomas Grant
Annual revenue: $28M
Employees: 145
Founded: 2011, incorporated in Delaware
Pending litigation: 2 matters (employment discrimination, contract dispute)
IP portfolio: 7 utility patents, 3 pending applications
Key customer concentration: top 3 clients = 62% of revenue
Outstanding debt: $4.1M revolving credit facility with First National Bank
Provide a preliminary risk assessment and identify areas needing deeper investigation.Expected Output:
- PII detected: company name, CEO name, bank name, financial figures
- Risk assessment: customer concentration risk, pending litigation exposure, IP valuation
- Deep-dive recommendations: litigation detail review, customer contract durability, patent claims analysis
Talking Points:
"Company name, CEO identity, bank relationship, revenue figures — all protected. The AI still delivered a structured risk assessment: 62% customer concentration is a red flag, two pending matters need litigation hold analysis, and the IP portfolio needs freedom-to-operate review. That's the output a partner expects from a second-year associate, delivered in seconds."
Chapter 4: Motion Drafting
Duration: 3 minutes Story: Jennifer drafts a Motion to Compel Discovery under FRCP Rule 37. The prompt includes case details, court information, judge name, and specific discovery failures.
Input:
Draft a Motion to Compel Discovery under Federal Rules of Civil Procedure Rule 37.
Case: Hartley v. National Insurance Group
Case No: 2024-CV-03847
Court: United States District Court, Northern District of Illinois
Judge: Hon. Catherine M. Reyes
Defendant has failed to produce documents responsive to Requests 4, 7, and 12 despite two meet-and-confer attempts on January 15 and February 3, 2025. The outstanding documents relate to internal claims handling procedures, adjuster training materials, and communications regarding the denied claim. Defendant asserts work product privilege but has not provided a privilege log.Expected Output:
- PII detected: party names, case number, judge name
- Properly formatted motion with caption, statement of facts, legal argument, and prayer for relief
- Rule 37 analysis: failure to produce, inadequate privilege assertion, no privilege log
Talking Points:
"Party names, case number, judge identity — all tokenized. But the motion itself is properly structured: caption, statement of facts, legal argument citing Rule 37(a)(3)(B), and a prayer for relief including fees. The AI even caught that Defendant hasn't produced a privilege log, which undermines the work product claim. That's substance, not just form."
Chapter 5: Ethical Wall Audit Trail
Duration: 2 minutes Story: Jennifer opens the admin console to review the audit trail — every matter access, ethical wall enforcement event, and compliance mapping for ABA Model Rules.
Input:
(No chat input — this chapter navigates Audit Log and Compliance admin panels)Expected Output:
- Audit Log: matter-scoped access entries, ethical wall enforcement events
- Compliance dashboard: ABA Model Rules 1.6, 1.7, 1.9 control mappings
- Per-matter access history with timestamps and user identity
Talking Points:
"Every matter access, every ethical wall enforcement, every query is logged with full attribution. When the disciplinary board asks how you maintained privilege across concurrent matters, this audit trail is your answer. It's not a policy document — it's proof."