A full-day, onsite working session with your partners. We find where AI is already in use, where it can take on real work, and the guardrails your clients and your profession expect. You leave with a policy, a plan, and a working first version, not a slide deck.
“I'm fairly sure our associates paste deal documents into ChatGPT. I just don't know which ones.”
Law firms
“If someone uploaded a client's return to a free AI tool, we'd never know.”
CPA firms
“What happens when an embargoed announcement ends up in a chatbot?”
Entertainment and PR
Most firms we meet are already using AI, just not on purpose. The workshop turns scattered, invisible use into approved tools, written rules, and one workflow that actually saves time, in a single day with the people who make the decisions.
We review a sample of your network's web (DNS) traffic to see which AI tools your team already uses. It shows the tools, never anyone's messages.
With your leaders in the room, we trace how work moves, from the first client call to the final invoice, and mark the routine work AI can take on.
We rank the opportunities by value and risk, pick the tools you'll approve, and draft the rules: what goes in, what never does, and who reviews the output.
We build a working first version of your top workflow on tools you approve, so you see it run before we leave.
Owners and dates for rolling out what we built, plus questions for your insurance broker about how your cyber and professional liability policies treat AI.
AI tools change every month. Our retainer vets new tools before your team adopts them, keeps your policy current, and confirms a person still reviews what AI produces. Priced monthly and quoted after the workshop. How Defensible AI works →
Everything below is included: the shadow AI review, a draft AI use policy ready for partner review, a ranked map of where AI fits in your firm (and where it doesn't), a working first version of one AI workflow, a 30/60/90 day plan with owners and dates, and questions for your broker about how your insurance treats AI.
A law firm, CPA or accounting firm, or entertainment, creative, or PR business with 5 to 100 people; already using AI without a written policy; fielding client questions, outside counsel guidelines, or contract clauses about AI; and able to get the people who make technology decisions in the room for the full day.
You want custom AI software or models built from scratch (we set up and govern tools, we don't develop them), you can't clear a full day with your decision makers, or you're after a slide deck instead of something that runs.
The guardrails are built around your duties of competence and confidentiality (ABA Model Rules 1.1 and 1.6): what goes into AI, what never does, and which attorney reviews the output. We map intake, conflicts, drafting, and work like first-pass summaries of talent and production agreements, and help you answer client and outside counsel questions about AI. We handle the technical side; your ethics counsel decides.
The rules center on safeguarding client tax and financial data under your professional standards and IRS safeguarding requirements. We trace engagements from document collection and organizers through prep, review, and client communication, and we schedule the day around your busy seasons.
The rules protect unreleased work, embargoed announcements, and client confidential material, and someone checks AI output for accuracy and voice. We trace projects from the brief and pitch through production, approvals, and wrap, and give you plain answers for client contracts and questionnaires that now ask about AI.
No. We review a sample of your network's web (DNS) traffic to see which AI tools are in use. It shows the tools, never anyone's messages, and you get the report before the workshop day.
Take the one-pager to your partners:
Law firms (PDF)CPA firms (PDF)Entertainment and PR (PDF)Other professional firms (PDF)Tell us about your firm and we'll find a day that works, including around busy season. The shadow AI review runs before the day, so we arrive knowing what's already in use.
