Engineering

EM: 1:1 prep + roadmap sanity check

Engineering managers' weekly preparation often gets squeezed. AI-assisted briefings make 1:1s and roadmap reviews substantively better.

Yash ShahApril 2, 20264 min read

An engineering manager described his Sunday-evening pattern: "I open my notes for tomorrow's 1:1s and realise I haven't kept up with what each report has been doing this week." The reports' work was visible — commits, PR reviews, project updates — but assembling it into 1:1-ready prep was its own chore. The chore got skipped most weeks. The 1:1s suffered.

Claude Code makes the prep automatic. The 1:1s become substantively better because the EM walks in informed.

The 1:1 prep brief

Before each 1:1, the AI assembles a brief on the report:

What they shipped this week. Commits, PRs merged, tickets closed. With links and brief context.
What they reviewed. PRs reviewed, with notable comments.
What they discussed. Slack messages in significant threads (with privacy considerations).
What they're working on now. Active tickets, in-progress branches, recent activity.
What's blocking. Tickets that aged, PRs awaiting review, dependencies surfaced in their work.
Recent feedback signals. From other team members, customer feedback if relevant, peer-review patterns.
Concern flags. Patterns that might warrant a manager's attention — sustained overtime, drop in activity, unusual escalations.

The EM reads the brief in 5 minutes. Walks into the 1:1 grounded.

The conversation, not the data

The brief isn't a script. It's a starting point:

"I noticed you spent a lot of this week unblocking the platform team — how's that landing for you?"
"The migration ticket has been open for three weeks; what's making it sticky?"
"Your last few PRs have been smaller than usual; intentional or context-switch fatigue?"

The conversation is what the EM is paid for. The brief makes the conversation grounded instead of generic.

Concern flags

The flags are surfaced gently:

Hours pattern (lots of weekend commits) → manager checks in on workload.
Sustained low PR-review participation → manager checks in on team-investment energy.
Unusual ticket-closure pattern (lots of small fixes, no feature work) → manager checks in on growth and challenge.

These aren't accusations. They're prompts. The manager decides what (if anything) to address.

Roadmap sanity check

A separate weekly brief: roadmap status:

Items on track.
Items at risk (with the indicators).
Items completed since last review.
New dependencies surfaced.
Resource issues.

The EM reviews. Surfaces concerns to leadership early. Adjusts plans before they slip publicly.

Follow-up discipline

The 1:1 produces follow-ups. The AI helps:

Drafting the manager's notes (action items, things to follow up on next week).
Surfacing whether last week's follow-ups were addressed.
Flagging recurring issues across multiple weeks.

This is the discipline that makes 1:1s feel valuable to reports. Without it, conversations don't seem to compound. With it, the report can see the manager paying attention.

A weekly ritual

A team running this for a quarter:

Manager prep time per 1:1: 5 minutes (brief read) + reflection.
Quality of conversations: measurably up (per manager and report self-report).
Concern catch rate: faster (manager spots patterns earlier).
Roadmap risk catch rate: faster.

The compounding effect: reports feel more managed, less ignored. Retention improves.

What stays human

The conversations.
The decisions about what to escalate.
The relationship work.
The career conversations.

Senior management work. The AI handles the data prep.

What we won't ship

Surveillance of individual employees beyond what's appropriate.

Performance-evaluation inputs from this brief. The brief is for the manager's prep, not for performance comp.

Automated escalations to higher management without the manager's decision.

Sharing 1:1 brief content with the report or anyone else.

How to start

Set up the brief for one report. Run it for two weeks. Refine. Expand to the rest of the team.

Close

1:1 prep with Claude Code is the difference between phoning it in and showing up grounded. Manager time is preserved. Conversations get richer. The reports feel managed in a way that compounds. Engineering management gets sharper across the team.