Before anything else: consent and confidentiality
Two structural properties of Earmark make this workflow safer than it might otherwise be. Both are worth knowing before you start:- Workspace isolation. Meeting content is private per user — your 1:1 artifacts are not visible to anyone else in your workspace unless you share them. That includes Chat: a teammate’s Chat never uses your meetings unless you add them to a shared project. Keep 1:1s out of shared projects. See Security and privacy.
- No training on your data. Earmark does not use your transcripts to train models.
What the artifact looks like
A worked example — a regular 1:1 partway through a quarter:The 1:1 template
Save it as a private template
1
Start in the Composer
There is no built-in 1:1 template, so the fastest path is to paste the prompt above into a new task on a real 1:1 and iterate in the Composer until the output lands.
2
Tune for your style
Some managers want a tighter Updates section; some want a more structured Career notes section. Adjust the headings to what you actually use. The shape matters more than the exact wording.
3
Save as Private, not Workspace
For 1:1s, save the template with Private visibility. Each manager runs their own 1:1s privately; the template does not need to be shared at the workspace level. See Custom templates.
Run it on a single 1:1
1
Confirm consent before the first recorded 1:1
In writing. Standing consent applies after that, but honor any “let’s go off-record” immediately — pause recording without friction.
2
Pre-seed the meeting
Add the saved 1:1 template as a task. For recurring 1:1s, attach it once to the calendar series. See Before a meeting.
3
Prep from past 1:1s
The day before, open Chat and ask “Help me prepare for my 1:1 with Priya tomorrow using our past 1:1s.” With your calendar connected, Chat pairs the upcoming 1:1 with your earlier ones. Ask it for the open threads and both sides’ commitments, and each comes back with a source. It’s the fastest way into the continuity loop below.
4
Use Customize context for continuity
Before the call, paste the Open threads and What we’re bringing to next 1:1 sections from last week’s artifact into the Customize context dialog. The template will use them to produce the Follow-up from last 1:1 section automatically.
5
Be present in the conversation
Don’t take notes manually. That’s the point. If something feels off-record and shouldn’t enter the artifact, say so — pause the recording and continue.
6
Three-minute cleanup
Within the hour:
- Move sensitive content out. If anything in the artifact is too sensitive to share with the report (a performance concern you’re forming, something confidential you mentioned in error), reframe it or move it to your private notes.
- Strip inferred emotional states. “Priya seemed frustrated about X” is the model interpreting. If she said “this is frustrating,” keep that as a quote. Otherwise, cut the editorial.
- Confirm the Open threads section. This is what carries to next week. Add anything the model missed. If you’re not sure whether a thread is new, click Chat in the meeting header and ask “Did this come up in the previous meeting?”
- Confirm commitments. Each side’s commitments are explicit, with deadlines.
7
Share with the report — same day
A Slack DM with the artifact pasted in, an email, or a link to a shared doc. Whichever the relationship has agreed on. Same-day delivery is what tells the report that the workflow is real and that their threads matter.
The continuity loop
The single highest-leverage habit, and the one most managers skip. The first 60 seconds of every 1:1 should be the manager reading the previous artifact’s Open threads out loud:- “Last time you brought up the review latency. Where are we?”
- “I said I’d think about your stretch project. Here’s what I’ve been thinking.”
- “We agreed to revisit the principal track conversation. Should we do that today?”
Variations
Same skeleton, three close relatives. Each is a saved private template.Skip-level
For senior leaders meeting with their direct reports’ reports. The shape is similar to the 1:1 but tilts toward signal worth flagging for the senior leader.Skip-level artifacts stay private to the senior leader. Do not share with the skip’s direct manager unless the skip explicitly consents. Skip-levels rely on confidentiality — using one to surface “feedback up the chain” without permission breaks the meeting type.
Candidate interview
For evaluative interviews. Designed to push into your ATS scorecard. For a full hiring loop — rubrics, screens, reference checks, calibration debriefs — use the hiring workflow instead; this section is the lighter version when interviews are one meeting type among several you run as a manager.Standup
For brief team-wide status meetings. Light touch — standups do not need elaborate artifacts.Working across many 1:1s
The biggest moves managers want — performance review prep, promotion case building, interview calibration across a hiring loop, disengagement detection across a quarter of 1:1s — all draw on many meetings. Ask Chat from the sidebar. It searches the meetings you recorded, cites each one, and ends with a Coverage line showing the date range and how many meetings it reviewed. Check that count against the 1:1s you actually had. Name the review period when you ask. Your chats are private to you.- Performance review prep. “Across my 1:1s with Priya since January, list her open threads, the commitments each of us made and whether they closed, and anything that touched on growth or feedback. Cite each meeting.” Then read the sources before anything goes into the review. The synthesis gives you the material; the assessment is yours.
- Promotion case building. Same pattern, oriented to the competencies at the target level. Ask for specific moments and her own words. The structure of the 1:1 template (specific moments, verbatim quotes, growth notes) is what makes those answers precise.
- Interview calibration across a loop. Each interviewer adds their interview to a shared project for the loop, and project chat synthesizes across them. See the hiring workflow.
- Disengagement detection. Ask about what was said, not how anyone felt: “Over my last quarter of 1:1s with Priya, which topics did she raise each time, and which threads have stayed open for more than three meetings?” Then read the artifacts it cites. A pattern is a reason for a conversation, not a conclusion. If you’d rather run your own analysis, export local transcript files for the period.
- Find a specific past 1:1 with the command menu (
Cmd+K/Ctrl+K). To find when a topic first came up, ask Chat, and say “ever” so it searches your full history: “When did Priya first raise the principal track, ever?”
Common pitfalls
- Recording without consent. Non-negotiable.
- Recording sensitive conversations that should not be recorded. Comp delivery, terminations, mental health discussions. Hold the line. Use temporary meetings for sessions you want to capture without keeping.
- Sharing the artifact selectively. The 1:1 artifact is a shared resource. Sometimes-sharing destroys trust faster than not-sharing at all. Pick one (always share) and commit.
- Private observations leaking into the shared artifact. The model has no way to know what is private. The cleanup is where you enforce the boundary.
- Skipping the continuity loop. Without reviewing open threads at the top of every 1:1, the workflow becomes a more efficient version of forgetfulness.
- Performance reviews written from recent memory anyway. If you have six months of artifacts and still write the review from the past three weeks, you have defeated the purpose. Ask Chat across the whole review period first.
- Adding 1:1s to a shared project. Everyone in a project can ask Chat about its meetings and see short excerpts. Keep 1:1s and skip-levels out of shared projects entirely; interviews go only in their own loop’s project.
- Vague interview scorecards. “Smart, good fit” is not interview signal. Force evidence-grounded language.
- Sharing skip-level signal with the skip’s manager. Unless explicitly consented to. Skip-levels rely on confidentiality.
- Letting the workflow become surveillance. The point is to support the relationship, not monitor the report. If you notice yourself using the artifacts for control rather than care, something has gone wrong.
- Forgetting that 1:1s are about presence. The artifact is downstream of the conversation. If the workflow makes the conversation feel mediated or transactional, the conversation has to win.
Where to go next
- Workflows — the general shape this is an instance of
- Chat — prep from past 1:1s and synthesis across a review period
- Temporary meetings — for sessions you want to capture without keeping
- Security and privacy — workspace isolation, no training on your data, retention controls
- Custom templates — visibility, sharing, and edit permissions
- Before a meeting — pre-seeding and using Customize context for prior-meeting carryover
- Composer — for tuning a prompt before saving as a template
- Action items workflow — for the extraction layer this workflow lightly uses
- Shareable summaries workflow — for the audience-aware sharing pattern (different in tone for internal 1:1s)
- Structured meeting notes workflow — the workhorse skeleton this workflow specializes for relationship meetings