Cullis AI consulting is a three-tier engagement model for Montana law firms. We do not sell speed. We sell defensibility. Every recommendation maps to ABA Model Rules 1.1, 1.6, and 5.3, and produces an evidentiary record the firm can stand behind in front of a malpractice carrier, the State Bar, or opposing counsel.
"Most legal-tech consultants sell speed. Cullis sells defensibility. Efficiency asks: did this make the lawyer faster? Defensibility asks: if a tribunal challenged this work product tomorrow, could the lawyer prove competent supervision?"
Every Cullis recommendation pushes the firm's AI posture in one direction: from vulnerable to defensible. The question is not whether the firm is using AI. The question is whether the firm can prove, after the fact, that the use was supervised, documented, and compliant.
AI in use without sanction, without supervision, and without record. The firm cannot answer basic questions from a client, an insurer, or the State Bar about what tools are touching client data, who authorized them, or what oversight is in place.
Every tool sanctioned. Every workflow supervised. Every output documented. When a tribunal, a malpractice carrier, or the State Bar asks how AI was used, the firm answers with paperwork, not memory. The evidentiary record is the product.
Most firms enter at Tier 1 and graduate based on what the audit uncovers. Each tier produces a discrete, defensible deliverable. You can stop at any tier; the prior tier's evidentiary record stands on its own.
"We need to close the malpractice gap immediately."
"Let us train your team to use the tools you already pay for, safely and defensibly."
"We will show you how to deploy AI in a way that increases firm profitability, not just productivity."
Every Cullis AI deliverable maps to one or more of these three Model Rules. These are not abstract aspirations. They are the framework a malpractice carrier, a State Bar grievance committee, or an expert witness will use when reviewing the firm's AI practices.
ABA Formal Opinion 512 made the implications for generative AI explicit: a lawyer must understand how the AI tool functions, what data it ingests, what it retains, and how its outputs may be inaccurate or biased. Technical incompetence is no longer a tenable defense.
Rule 1.6(c) requires lawyers to make reasonable efforts to prevent unauthorized disclosure of client information. Pasting a client's will into a free public chatbot is the modern equivalent of leaving the firm's filing cabinet on the sidewalk.
Rule 5.3 was written for paralegals but applies with equal force to AI agents. The lawyer who deploys an AI tool is supervising a nonlawyer. That supervision must be documented. The Human-in-the-Loop workflow is, at its core, a Rule 5.3 compliance mechanism.
Every Cullis engagement begins with a structured intake. No surprises, no high-pressure sales. The objective of Phase 0 is to give you a written picture of where your professional duties and your technology stack may be misaligned.
Investments at every tier are calibrated to the engagement scope and the firm's complexity. Reach out to start a conflict check and schedule your discovery call.