17 August 2026
A follow-up email that gets acted on has four things in it: what was decided, who owns each action, the date it's due, and what's still unresolved. Everything else — the discussion that led there, the tangents, the pleasantries — belongs in the meeting notes, not the email. If a recipient has to scroll to find their name, the email has already failed.
Decisions come first, stated as facts, not as a recap of the debate. 'We're moving the launch to the 14th' is a decision. 'We talked about timing and there were some concerns about the 14th but also reasons to keep it' is not — it's a transcript pretending to be a summary, and it forces the reader to figure out what actually happened.
Owners go right next to the decision they belong to, not in a separate list at the bottom. If someone has to cross-reference two sections to find out it's their task, most won't bother. Deadlines attach to the owner the same way — a task without a date is a suggestion, and suggestions get deprioritized the moment something louder shows up.
Open items go last, and they should be genuinely open — questions nobody answered, decisions punted to next time, dependencies waiting on another team. This is the section people skip if they're only checking their own action items, so it needs to be short enough to skim in one pass.
Pasting the transcript, or even a long AI-generated summary of it, moves the work of extracting decisions from the writer to every single reader. Ten people now each spend two minutes hunting for the one line that applies to them, instead of one person spending five minutes writing four lines that apply to everyone.
A transcript also has no hierarchy. A throwaway comment and a hard commitment read the same on the page, so the reader has no signal about what actually matters versus what was just someone thinking out loud. The result is that people either skim past the real decision because it's surrounded by noise, or they reply asking 'so what do I actually need to do' — which means the email did not do its job and someone has to redo it in a thread.
The instinct to send everything usually comes from wanting to be transparent or cover yourself if something gets disputed later. That's a real need, but the fix is a link to the full notes or recording, not a wall of text in the body of the email. Transparency and readability are different problems and they need different formats.
AI-generated meeting summaries have gotten good at capturing everything that was said, which is exactly the problem — capturing everything and summarizing well are not the same skill. A summary that includes small talk, restated agenda items, and every attendee's individual comment on a topic is longer and no more useful than the transcript it was supposed to replace.
Verbatim quotes should only appear when the exact wording matters — a legal commitment, a specific number, a promise made to a client. Otherwise paraphrase; quoting someone's exact phrasing on a routine point adds length without adding information.
Cut anything that was raised and then dropped without a decision or an owner. If three people floated ideas and the group moved on, that's not an action item and it's not an open question either — it's just discussion, and discussion belongs in the notes archive, not the email that lands in someone's inbox at 5pm.
A follow-up email that follows this shape rarely needs more than eight to ten lines total, regardless of how long the meeting ran.
AVAY's AI participant keeps a running log of decisions and owners while the call is happening, so by the time the meeting ends there's already a structured record to draw the follow-up email from rather than a transcript to mine after the fact. That closes most of the gap between 'meeting ended' and 'email sent.'
It doesn't remove the judgment call of what to cut, though. If a file was shared or pinned live during the call, it's visible to people in that meeting but isn't part of the persistent notes afterward — so if the follow-up email references a document someone pulled up mid-call, attach it directly rather than assuming a link to the meeting record will surface it. The email is still the place where someone decides what a stranger to the call actually needs to see.
| Full transcript or long summary | Four-line structured email | |
|---|---|---|
| Finding their action item | Scrolls or searches for their own name | It's in the first few lines, grouped by owner |
| What stands out | Every comment reads with equal weight | Only decisions and actions are stated as facts |
| Likely reply | A reply asking what they're actually supposed to do | A date, a yes, or silence because it's already clear |
Long enough to state the decisions, owners, deadlines, and open items — usually eight to ten lines regardless of how long the meeting ran. If it's longer than that, it's carrying discussion that belongs in the linked notes instead.
Link to it rather than pasting it into the body. A link lets the few people who genuinely need the context go get it without forcing everyone else to scroll past it to find their own action item.
Say that directly — 'no decisions were made, here's what we're still weighing' is more useful than forcing a structure that doesn't exist. List the open questions and who's expected to bring an answer to the next meeting.
Whoever ran the meeting or owns the outcome, not whoever happened to take notes. The sender's name attached to the email carries an implicit signal about who's accountable for the actions landing.
Same day, ideally within an hour or two while the context is still fresh for everyone who attended. A follow-up sent two days later gets read with less attention and generates more 'wait, what was this about' replies.
A follow-up email works when it states decisions as facts, puts an owner and a date on every action, and lists what's still open — everything else, including the transcript, belongs in a link, not the body.
Meetings that take their own notes, in the browser: avay.ai.