The junior associate reads the agreement, identifies the standard risk provisions, flags the non-standard ones, and produces a memo. This takes 3 hours. You pay $1,500–$1,950. Claude reads the same agreement in 4 minutes. Identifies every risk provision. Flags every non-standard clause. Explains each in plain language. Suggests negotiation positions. Costs: approximately $0.04. In many cases, the AI analysis is more thorough — because AI doesn't miss clauses due to fatigue or rush, doesn't skip sections because they look standard, and doesn't bill for the time it takes to look things up.
This is the Fog Machine running in the legal profession — and the AI off-switch is already built.
What Legal Functions AI Handles at High Quality
Standard Contract Review: NDAs, vendor agreements, service agreements, consulting agreements, licensing agreements, employment contracts, independent contractor agreements — all standard commercial contracts follow predictable structures. AI reads them accurately and completely. Ask: "Summarize the key obligations I'm taking on under this agreement," "What are the exit provisions and under what conditions do they apply?" "Flag any indemnification clauses and explain the scope," "What intellectual property rights am I assigning?" "What are the limitations of liability and how do they compare to market standard?" "Are there any automatic renewal clauses?" Claude Opus, GPT-4o, or legal-specific AI tools (Harvey, Spellbook) produce accurate, complete answers to all of these questions from any uploaded document.
Policy Drafting: Employee handbooks, HR policies, remote work policies, IT security policies, data retention policies, confidentiality policies — all follow standard frameworks that AI executes with high accuracy. The AI draft typically matches or exceeds the quality of what a junior attorney would produce, because the policy frameworks are well-documented and AI doesn't introduce idiosyncratic stylistic choices that make policies harder to follow. Process: describe your company size, jurisdiction, and specific policy need. Ask AI to draft. Have local counsel review specific jurisdictional questions (30 minutes of attorney time, not 5 hours).
Legal Research: Identifying applicable statutes, regulations, and case law for standard legal questions is AI-executable. Perplexity Pro and Claude handle legal research questions with source citations. When to use: understanding your contractual rights before a dispute, researching regulatory requirements for a new business activity, identifying which compliance requirements apply to a given situation. When not to use: actual legal strategy in contested situations, novel legal questions without clear precedent.
Compliance Documentation: Terms of service, privacy policies, cookie policies, GDPR compliance documentation, CCPA notices, accessibility statements — all standard compliance documents follow regulatory templates that AI executes accurately. Have any compliance documentation reviewed by counsel in your jurisdiction before publishing — 30–60 minutes of attorney time. You're paying for the review, not the production.
What Legal Functions Still Require a Human Attorney
Actual Litigation Strategy: If you are in or approaching contested litigation, you need counsel. Not because AI can't analyze case law — it can. Because litigation strategy involves relationship knowledge (how this judge rules, how this counterparty's counsel behaves, what the jury pool looks like in this jurisdiction) that is not available in text form.
Novel Legal Interpretation: Situations with no clear precedent — where the legal answer genuinely depends on how a court interprets a new fact pattern against existing law — require attorney judgment. AI can research the existing framework; the interpretation of how that framework applies is legal judgment.
High-Stakes Negotiation: Contract negotiation in high-stakes deals involves relationship management, read-the-room capability, and tactical judgment about when to hold and when to concede. These are human skills.
High-Stakes Drafting: AI produces excellent first drafts of standard agreements. When the agreement is non-standard — complex corporate transactions, novel IP arrangements, high-stakes acquisition documents — attorney expertise produces more reliable results. Use AI for the first draft; use counsel for the specific high-stakes provisions.
The question is never "zero risk." The question is "acceptable risk for the cost." AI contract review at current capability levels is appropriate for standard commercial agreements. For those documents, the risk profile of AI review is comparable to attorney review at the junior associate level — at a fraction of the cost. Use attorney review for the documents where the stakes justify it. Apply the same cost-benefit analysis you apply to everything else in your business.
The Legal Stack Configuration
The full AI legal stack: Claude Opus for standard contract review ($20/mo, replaces $2,000–$8,000 in routine review), GPT-4o for NDA production ($0, on-demand), Claude + Notion AI for policy drafting ($20/mo, replaces $500–$2,000/mo in policy work), Perplexity Pro for legal research ($20/mo, replaces $500–$3,000/mo in research), Claude for compliance docs ($0, templates built once). Total: $60/month plus occasional episodic counsel. Replaces: $2,000–$10,000/month in standard legal retainers.
What you retain counsel for: high-stakes transactions, active litigation, novel legal questions. These are episodic, not monthly retainers. Pay for counsel when you need counsel. Stop paying for counsel to review contracts that AI handles accurately.
How to Transition Your Legal Relationship
Step 1: Audit your current legal spend. List every matter your attorney handled in the last 6 months. Categorize each as execution (what AI now handles) or judgment (what requires attorney expertise).
Step 2: Migrate execution functions immediately. Set up Claude or your preferred AI tool for contract review. Test it on your next incoming contract before sending it to your attorney. Compare the outputs. You will find the AI analysis is thorough — and frequently more comprehensive than the attorney's mark-up of standard provisions.
Step 3: Restructure the attorney relationship. Instead of a monthly retainer, move to project-based or hourly engagement for specific judgment-layer matters. Be explicit: "I'm handling standard contract review internally now. I'll engage you for [specific situation types] on a project basis."
Step 4: Build your standard document library. Use AI to produce your standard NDAs, contractor agreements, MSAs, and service agreements. Have an attorney review and approve the templates once. Then use the templates going forward without per-document legal review.
The complete Consultant Killer Stack — including legal, HR, strategy, and financial configurations — is in Obsolete By Noon. The $650/hour is for the brand on the letterhead. The work has been replaceable for two years.
Get the Full Stack on Amazon →Frequently Asked Questions
AI handles standard contract review (NDAs, vendor agreements, employment contracts), policy drafting (employee handbooks, HR policies, IT security and data retention policies), legal research for standard questions with source citations, and compliance documentation (terms of service, privacy policies, GDPR, CCPA notices). Total monthly cost: $60. Replaces $2,000–$10,000/month in standard legal retainers.
Human attorneys are required for: actual litigation strategy (involves relationship knowledge about how this judge rules and how counterparty counsel behaves), novel legal interpretation (new fact patterns against existing law), high-stakes negotiation (requires relationship management and read-the-room capability), and non-standard high-stakes drafting (complex corporate transactions, novel IP arrangements, acquisition documents). Use AI for the first draft; retain counsel for the specific high-stakes provisions.
The question is never zero risk — it is acceptable risk for the cost. AI contract review at current capability levels is appropriate for standard commercial agreements. For those documents, the risk profile is comparable to junior associate review at a fraction of the cost. AI doesn't miss clauses due to fatigue, doesn't skip sections, and doesn't bill for look-up time. Use attorney review for documents where stakes justify it; apply cost-benefit analysis consistently.
Move from monthly retainer to project-based engagement for specific judgment-layer matters. Be explicit with your attorney: "I'm handling standard contract review internally now. I'll engage you for [specific situation types] on a project basis." Build your standard document library using AI — have an attorney approve templates once, then use them without per-document review. This restructures the billing relationship from "everything" to "judgment only."
Claude Opus for standard contract review — upload any contract and ask specific questions about obligations, exit provisions, indemnification scope, IP assignments, and liability limitations. Harvey AI or Spellbook for legal-specific contract analysis with legal reasoning training. Perplexity Pro for legal research with source citations. GPT-4o for NDA and policy drafting. The combination covers 70–80% of what most organizations route to legal retainers.