A few months back, I was sitting with Evan — he runs a coworking space and had been doing a series of AI workflow sessions with me. We were working through how to build a system that kept his strategic projects current without anyone spending time manually maintaining status docs.
The problem he described is universal. He had project documents. He had meeting notes. He had email threads. And none of them reflected what was actually happening. The real status of any given initiative existed somewhere between three Slack conversations, a Granola transcript, and two emails he hadn’t replied to yet.
His team could only know the true state of things by asking him. And asking him meant interrupting him.
The system we designed to fix this is something I call the Master Memo.
What a Master Memo Is
A Master Memo is a living project document maintained automatically by an AI agent.
The agent reads incoming emails, meeting transcripts, and voice notes related to a specific project. After each new input, it updates the document — surfacing key developments, things team members have committed to doing, items that are currently stuck, and follow-ups that were promised but never closed.
The raw quote I wrote when we were designing it: “The purpose of the memo: information sharing, prompting action, tasks stuck, things we said we’re going to do but haven’t done. Following up on the previous week — you said you were going to do this, is this done?”
Nobody has to manually maintain it. The agent does it. The document just stays current.
Why Most Project Docs Don’t Work
The pattern I see across teams of every size: a project document gets created with good intentions. It has the initial scope, the stakeholders, the timeline. And then it becomes a fossil.
The meetings keep happening. The decisions keep getting made. But none of it makes it back into the document because updating the document is extra work — it’s overhead that competes with everything else on the agenda. So everyone stops trusting it. And the real source of truth becomes “ask someone who was in the room.”
The problem with that isn’t just inefficiency. It’s that the institutional memory lives in people instead of systems. When someone is out sick, or leaves, or just can’t be reached, a critical piece of project context goes with them.
How the System Works
The implementation isn’t complicated, but it does require setting it up correctly.
Step 1: Define the memo structure. What does the document need to track? For most project memos, this is: current status, key decisions made, outstanding commitments (with owner), things stuck, and open questions. This structure becomes the template the agent writes to.
Step 2: Connect the agent to your inputs. The agent needs to read emails related to the project (filtered by project name, participants, or a keyword), meeting transcripts, and any voice notes or async updates. In practice, this usually means connecting to Gmail (filtered), a Granola or Otter integration, and possibly a Slack channel.
Step 3: Set the update cadence. The agent runs on a schedule — daily or weekly depending on project velocity. It reads new inputs since the last run, decides what’s worth adding or updating, and writes directly to the shared document.
The version I showed one of my AI consulting clients earlier this year gave him a broader context file system — folders covering identity, operating style, and current projects, all built by feeding Claude a year of meeting transcripts and having it extract patterns. The setup took about an hour. Now it powers dozens of agents that reference the same information.
His takeaway: the leverage doesn’t come from having the files. It comes from agents using them.
What This Changes
The most obvious change is time. A team that was spending 30-60 minutes per week on manual status updates gets that time back. But that’s not really the point.
The deeper change is trust. When people know there’s a document that actually reflects reality — not what someone typed in three weeks ago — they reference it. They rely on it. The coordination overhead that comes from “does anyone know the latest on X?” drops significantly.
And the document itself becomes a useful object for retrospectives, onboarding new team members, or quickly briefing stakeholders who weren’t in the room.
Most meetings generate more follow-up work than they eliminate. A memo that writes itself changes that ratio.
To explore how to build AI systems like this for your own work, the 4-Day AI Sprint walks through how to design and deploy AI workflows that handle your most time-consuming recurring tasks.
Recommended for you
Want the full system? 25X is the flagship productivity system we teach.
Explore 25X →