How to Talk to Your Board About AI (Without the Buzzwords)

You want to implement AI. But your board is skeptical. Or they want a detailed plan before approving budget. Or they ask questions like "What's an LLM?" and you don't know how to respond without sounding like you're speaking Klingon.

Here's how to communicate about AI to non-technical stakeholders in a way that actually moves decisions.

What NOT to Do

Don't use jargon. "We're going to use large language models and machine learning algorithms to optimize workflow efficiency." Nobody cares and nobody understands.

Don't lead with technology. "ChatGPT uses transformers and attention mechanisms..." Boring and irrelevant.

Don't promise the moon. "AI will double our productivity!" You'll be wrong and you'll lose credibility.

Don't avoid the concerns. If your board is worried about privacy or liability, address it head-on. Don't pretend the risk doesn't exist.

What to Do Instead

1. Start with the Problem, Not the Solution

Lead with the business problem: "We're spending $200K per year on email management and document filing. This is non-billable, non-differentiating work that our people could be spending time on client relationships or strategic work."

Then introduce the solution: "AI can automate 60-70% of that work, freeing up time for billable work."

2. Speak in Terms of ROI, Not Features

Not: "We'll use natural language processing to categorize documents."

Instead: "Documents that now take a person 30 minutes to organize will be pre-organized by AI in 5 minutes. That's 2-3 hours per person per week back into billable time."

At $200/hour billing rate × 2.5 hours × 20 people × 50 weeks = $500K in recovered billable time per year.

3. Address the Risks Directly

Don't hide from concerns. Acknowledge them:

"Some people worry about AI making mistakes. Fair point. We're implementing AI to assist human judgment, not replace it. A person still reviews AI output. AI just does the routine parts faster."

"Some worry about data privacy. We're using tools with SOC 2 compliance and contractual protections. We're not uploading client data to third-party systems."

4. Show Real Pilots, Not Projections

Don't present theoretical models. Run a pilot and bring data:

"We piloted email triage with 5 people for four weeks. On average, each person recovered 3 hours per week. Cost of the pilot: $2,000. Value in recovered billable time: $12,000. ROI: 6x in month one."

5. Propose Incremental Expansion, Not Massive Transformation

Not: "We're going to AI-transform our entire firm."

Instead: "We're going to automate email management. Then we'll evaluate meeting summaries. Then document drafting. Each one is a separate decision based on results."

This feels less risky and is actually more likely to succeed.

The Three-Slide Pitch

If you have 10 minutes with your board, here's what to present:

Slide 1: The Problem

"Our team spends 4 hours per person per week on non-billable, non-differentiating work (email, document management, meeting notes). This costs us $X annually. It reduces billability and prevents us from focusing on client work."

Slide 2: The Pilot Results

"We piloted AI-assisted email management with 5 people. Results: 3 hours per person per week of recovered time, zero errors in categorization, 100% user adoption. Cost: $2K. Value: $12K. ROI: 6x."

Slide 3: The Plan

"We propose rolling out email management to our full team (investment: $10K, recovery: $40K annually). Then we evaluate the next use case. Each expansion is based on results."

How to Handle Objections

"Will this replace people?"

"No. We're using AI to eliminate boring work and give people time for meaningful work. We're not eliminating positions; we're changing what positions do."

"What if it makes mistakes?"

"AI assists human judgment, not replaces it. Someone always reviews the output. AI just does the routine parts faster."

"How much will it cost?"

"For email management: $15K per year in software and implementation. Returns: $40K per year in recovered billable time. Payback: 4.5 months."

"Aren't there privacy/security concerns?"

"Yes, and we're taking them seriously. We're using tools with enterprise security certifications. We're not uploading sensitive data to third parties. We have strong contractual protections."

The Key Principle

Talk about business outcomes, not technology. Talk about problems you're solving, not features you're implementing. Talk about data from pilots, not projections from vendors.

Your board will respond to that. And you'll get approval to move forward.