Beyond the Summary: AI, Institutional Memory, and Better Board Preparation

Key Takeaways

• 92% of board directors now use AI for board work, most often to summarize board books, and mostly with general-purpose tools like ChatGPT and Copilot rather than their board platform.
• Summaries make board preparation faster, not better. The bigger opportunity is using the board’s institutional memory to ask better questions.
• Any AI tool that touches board materials should meet four standards: approved information handling, permissions that follow governance, outputs you can review, and records the board can govern.

The board book lands about a week before the meeting. For a public system, it can be enormous. At the University of Louisiana System, board packets can run up to 800 pages, according to Dr. Claire Norris, the system’s senior vice president for advancement, research, and strategy, in an earlier AGB OnBoard conversation on AI readiness. For a smaller private college, it might be a third of that, but it still covers enrollment projections, the operating budget, an audit committee report, an accreditation update, and a tuition recommendation that needs a vote.

Trustees read it in pieces: on a flight, between meetings, late the night before. And more of them now hand it to AI first.

The Shift Has Already Happened

In OnBoard’s 2026 Board Effectiveness Survey, which collected responses from 531 board leaders across sectors, including education, through June 2026, 92% of board directors said they use AI for their board work, up from 69% a year earlier. The most common use was summarizing board books and reports. Only 5% said they use AI built into their board management platform.

Board professionals in this fall’s AGB OnBoard seminar series said the same thing. Joshua Merchant, chief of staff and secretary to the board of trustees at Oakland University, described the worry plainly: trustees putting financial statements, treasurer’s reports, or internal audits from the board book into a personal AI account. “My fear is that that could be happening,” he said. “It probably is, and I just don’t know about it.”

So the question for most boards is no longer whether trustees will use AI to prepare. It’s whether that use makes the board better at its job.

Faster Isn’t the Same as Better

Merchant named the risk better than any survey could. AGB OnBoard has spent more than a decade asking trustees to take their consequential role seriously, and that means getting into the details. A summary, he said, might give trustees “the genesis of the issues, but not necessarily the details,” and that gives him pause when they’re voting on something significant.

The survey points the same way. Half of respondents said AI made them more efficient at preparing materials. Far fewer pointed to better director preparedness (22%) or more engagement in meetings (7%).

The point of preparation was never to finish the book sooner. It’s to walk into the room ready to ask the question that changes the conversation. Heather Fehn, vice president, chief of staff, and secretary to the board of trustees at The College of New Jersey, has built her board’s preparation around exactly that. Before each committee meeting, trustees get a framing memo: what’s on the agenda, what they need to read closely, what action items are headed to the full board, and what questions they should be ready to engage with. AI now helps draft those memos, but the memo’s job hasn’t changed.

That’s the standard AI should be held to. A generic summary compresses what the administration already wrote. A better question draws on the board’s institutional memory: what it decided before, why, and what it asked to see next time. How has our discount rate moved over the last three years, and what has it done to net tuition revenue? When did we last raise the endowment spending rate, and why? Who objected? Questions like these keep trustees where Fehn wants them, “at the 35,000-foot level,” focused on strategy and oversight rather than operations. For more on where questions like these come up in a higher education board’s work, see our guide, Using AI in the Higher Education Boardroom.

That memory is also the hardest thing for a board to hold onto. Trustees rotate off, presidents and cabinet members change, and the people who remember why a decision was made aren’t always in the room when it comes back.

What Good AI Preparation Looks Like

The first session of the seminar series offered a useful lens: don’t just judge the output, judge the workflow. Applied to board preparation, that means four things.

Approved information handling. Board materials go only into tools the institution has approved. Many campuses already license enterprise AI; Oakland and TCNJ are both Google campuses with their own Gemini environment. The gap is trustees using personal accounts. TCNJ is moving its trustees to institutional email accounts, in part so they can use the campus’s approved tools.

Permissions that follow governance. Executive session materials, personnel matters, and committee-only documents need to stay that way. An AI tool should see only what the person asking is allowed to see.

Outputs you can review. Answers should point back to the source document so a trustee can check them in seconds. One first-year trustee at a private college noted that asking AI the same question on different days can produce different answers. Sources are how a board catches that. At one private university, the plan is for cabinet members to review and sign off on AI-drafted committee summaries before trustees see them.

Records the board can govern. Confidential material should stay inside a secure board system, under the institution’s own retention rules, and should never be used to train AI.

Behind all four is the point every panelist came back to. AI can surface the history and the pattern. Deciding what it means is still the board’s job. As Fehn put it, “There’s just nothing that replaces human judgment.”

Where Board Professionals Can Start

Board professionals are usually the ones asked to write the policy, and most haven’t yet. In the survey, 38% of boards had no AI policy at all and only 6% had one that was enforced. None of the institutions that spoke about it during the series had a board-specific AI policy in place.

They did have practical starting points:

• Ask trustees how they already use AI and what training they want. One private university surveyed its board.
• Work with general counsel to add language on AI use with board materials to the annual conflict of interest statement. Oakland is considering this.
• Start with guidelines rather than a formal policy if that fits your board’s culture.
• Label AI-assisted summaries and have a named leader review them before they’re shared.
• Give trustees approved tools, so the secure option is also the easy one.

Where AGB OnBoard Fits

This is the problem we built AI Assist to address. Inside AGB OnBoard, trustees can ask questions of their current and past board books and meeting materials in plain language, so the board’s institutional memory is available to every trustee, not just the longest-serving ones. Answers cite their sources, respect each user’s existing permissions, and stay inside the board portal. Your data is never used to train AI.

Four Questions to Bring to Your Next Board Meeting

1. How are our trustees already using AI to prepare, and with which accounts?
2. Do we know where our board materials go when they’re pasted into those tools?
3. When a trustee wants to know what the board decided before, how long does it take to find out?
4. Is our preparation producing better questions in the room, or just shorter reading time?

The fourth question matters most. AI is already part of how your trustees prepare. The opportunity is to make it count in the boardroom.

Looking for more ways to use AI in the higher education boardroom? Our guide, Using AI in the Higher Education Boardroom, walks through the moments in a board’s work where AI makes the biggest difference, from preparing for the next meeting to finding what the board decided before.