Write quarterly performance reviews for my team
Your AI CEO writes the reviews you keep pushing to next week.
You'd think this needs a clear Saturday and a strong coffee — Astra hands you 8 drafts before lunch.
The short answer
Astra writes performance reviews by reading each report's actual work output and turning it into a balanced, specific draft you can polish in 10 minutes. For each direct report she pulls Linear (issues shipped, time per ticket), GitHub (PRs reviewed and authored, code quality signals), Google Calendar (meeting load, focus time), 1:1 notes from your Notion or Lattice, and any 360 feedback collected. She drafts a review structured as: top 3 wins with specific examples, top 2 growth areas with concrete next steps, alignment with role expectations, and one stretch question for the conversation. The draft is honest — neither sycophantic nor harsh — and grounded in observable evidence. Output: one Lark draft per person, queued for your 10-min edit pass. A 6-person team that used to eat your weekend now takes 60 minutes total.
How Astra actually does it
- 1Pull work output per person
Linear (issues, cycle time), GitHub (PRs authored/reviewed, lines changed), calendar (meeting load), 1:1 notes from Notion/Lattice, peer 360 if available.
LinearGitHubNotionLattice - 2Identify wins and growth areas
3 wins per person with specific examples (project name, impact, behavior). 2 growth areas with concrete next-quarter actions, not generic advice.
- 3Draft balanced review
Structure: wins, growth areas, role expectation alignment, stretch question. Tone: honest, specific, neither sycophantic nor harsh.
- 4Cross-check against role rubric
Compare to your role expectations doc (or industry-standard SDR/IC/Manager rubric if you don't have one). Flag any expectation gaps.
- 5Queue in Lark for edit
One draft per person in Lark. Inline edits supported. Once you approve, ships to Lattice/Notion as the official review record.
LarkLatticeNotion
What it looks like in chat
One Lark draft per direct report — wins with specific examples, growth areas with concrete next steps, role-rubric alignment, and a stretch question — ready for your 10-min edit and ship to Lattice.
Quarterly per review cycle.
Ask Astra this right now
We'll spin up your workspace, hand the prompt to Astra, and you see the answer in 60 seconds. Free.
Try this with AstraFrequently asked questions
What if I have non-engineers on the team?
Astra adapts the data sources: for sales reps she pulls Salesforce (closed-won, pipeline coverage, activity); for designers Figma (files shipped, comment threads); for marketers GA4 + content output. She'll ask if she's not sure which sources matter for a given role.
Will the reviews sound generic and AI-written?
No — she grounds every claim in a specific example (PR number, project name, observed behavior). Generic praise ("great team player") is replaced with specific evidence ("Lattice peer feedback from 4/5 reports cited her unblocking them on the migration"). The 10-min edit pass adds your voice.
What if I haven't kept 1:1 notes?
She works with what's there. Without 1:1 notes she leans more heavily on observable work output (Linear, GitHub) and skip-level feedback if available. She'll explicitly note what data was missing so you can supplement from memory in the edit pass.
Can she also rate or rank performance?
She maps each person to your role rubric ("meeting bar / exceeding bar / below bar") with reasoning. If you use a numeric calibration (like Lattice 1-5), she'll propose a score with justification — but final calibration stays with you and your leadership team. She never publishes ratings without your sign-off.
Run your one-person company.
Hire your AI team in 30 seconds. Start for free.
Free to start · No credit card required · Set up in 30 seconds