Notes: Choose what matters when everything feels urgent

Elena's morning. In this prepared example, Elena manages a department, and her morning opens with a screen full of new requests.

Two of them look very different:

• an internal formatting request, stamped urgent and due at noon
• a short, polite message from a peer director asking to borrow Dev, a senior engineer, for 3 days

The red stamp pulls her eye first. Yet giving Dev away for 3 days would quietly delay a software release her team has promised a client for Friday.

What an agent can and cannot see. An agent can count 7 messages and notice that 2 are marked urgent. It cannot work out from that count what breaking Friday's promise would cost the client, or her team.

So the help worth having here is a sorted view that puts the request about Dev in front of Elena with the Friday release attached, so she thinks about it before she replies. Five more summaries would only add to the pile.

Elena has 4 items waiting:

1. a peer director asking to borrow Dev for 3 days
2. a formatting request stamped urgent for noon
3. a client progress update, drafted and ready to review
4. a supplier pricing confirmation, waiting on the vendor's finance team

If she works down her inbox by arrival time, her best hour of the morning goes on the formatting request and the supplier thread. By the time she reaches Dev, the other director may already have written Dev into his plan.

Sorting into 3 groups. Elena gives the agent the 4 requests and her department's project records, and asks it to separate decisions that need her, drafts ready to inspect, and items waiting on someone else. The supplier thread goes to waiting, because nothing Elena does moves it until the vendor replies. The client update goes to drafts, for her afternoon review.

Correcting the order. Two decisions are left, and the agent first puts the formatting request on top because noon is the earliest time. Elena tells it what it could not know: Dev leads the engineering for Friday's committed client release, and a late formatting decision affects nobody outside the department.

Then she gives it her ranking rule, in this order:

1. the consequence of delay
2. how important the commitment is
3. the actual deadlines
4. what other work depends on the choice

With that rule, Dev's request moves to number 1, with a note from the project record about the Friday release.

One saved page, not a fresh list each morning. The queue lives as one page in her department project. At each review point the agent updates it and brings back only what needs Elena, and a change that threatens a committed date jumps to the top.

The situation. Maya, the proposals director from the previous section, is answering an RFP from Meridian Manufacturing for a 6-week data migration. Her only 2 senior integration specialists are booked on another client's launch, planned to run into weeks 3 and 4.

Her adapted skill already checks specialist time rather than total headcount, so the agent compares 3 options:

1. bid on the full 6-week scope
2. propose a narrower phased scope that moves integration to week 5
3. decline to bid

Asking for the assumption. Maya does not stop at the option that looks best on paper. She asks the agent for the pivotal assumption behind the full-scope bid.

The answer is plain: the full bid works only if the other launch finishes early, by day 10 at the end of week 2, without a single day of spillover. Only then are both specialists free for Meridian in weeks 3 and 4.

Weighing it. That launch depends on a third-party vendor whose testing schedule is still unconfirmed. Betting the whole Meridian contract on day 10 is too risky, so Maya chooses the phased scope. She accepts a smaller first contract to protect the launch her team already promised.

She also names what would reopen full scope:

• written confirmation that the current launch has passed final testing
• early enough to free both specialists by day 10
• arriving before the bid is due on Friday, October 9, at 5 p.m.

Thursday's question. The day before the bid is due, a commercial vice president asks whether the team can expand to the full 6-week scope. If Maya's reasoning is buried in a 30-message chat, rebuilding it is slow and easy to get wrong, and a new chat will not remember the old one on its own.

So right after confirming her choice, Maya asks the agent for a concise decision record with 5 items:

1. the confirmed choice
2. the reasons behind it
3. the alternatives considered
4. the fact still to confirm
5. the condition that reopens full scope

When the draft comes back, she checks the 2 things automated summaries often get wrong:

1. the record names the narrower phased scope as the final choice, even though the chat spent paragraphs on the full-scope bid
2. the week-5 specialist hours are still listed as a fact to confirm before contract signature, not as settled

This draft gets both right, so she approves it, and the agent saves it in the Meridian Proposal project folder beside the RFP summary and the capacity note.

Checking it can be found. On Thursday, Maya opens a fresh conversation inside the same project and asks the agent to retrieve the record. It points to the saved file and reports both open items: the week-5 specialist hours, and the confirmation before Friday's deadline that would reopen full scope.

If a chat has no access to the project folder, the right answer is that the record is inaccessible, not a guess at the reasons from memory.

Elena, before and after. Her first suggested order put a routine formatting request on top because of its time stamp. Now she can say she is handling Dev's request first because lending Dev for 3 days would put Friday's committed client release at risk.

Maya's explanation from her saved record has 4 parts:

1. the confirmed choice: propose the narrower phased scope for Meridian
2. the protected commitment and accepted tradeoff: protect an existing client delivery, accept a smaller first contract
3. the fact still to confirm: the week-5 specialist hours, with the engineering lead
4. the condition that reopens full scope: confirmation before Friday's bid deadline that the active launch finished testing early

Some leaders worry that admitting an open uncertainty makes a decision sound weak. In practice it does the opposite, because colleagues know what to watch for instead of discovering the risk halfway through delivery.

Halfway through the course, you can now:

• hand an agent your meeting follow-through
• keep your background and methods in a project and a skill
• sort your week with an attention queue
• keep the reasons for a choice in a decision record

A decision sitting in a project folder only changes the work once the people around you understand what it means for them. Section 4 is about drafting those messages so they still sound like you.