Mid-year reviews w/ Claude

Comment MIDYEAR and I'll send them over

In most companies I talk to, HR leads AI adoption.

Mid-year reviews are the perfect place to show it. And the timing is right now.

Here’s the problem every cycle has: the writing. A manager with 6 direct reports sits down to write. Six months of work per person, and all they have is memory - which really means the last three weeks. About an hour per review, half of it staring at a blank page. Employees do the same with their self-reviews.

Meanwhile the context already exists. Managers run Granola or Fathom in their 1-1s. Decisions live in Slack and email. Goals and action from previous meetings sit in Google Docs. Six months of evidence, captured automatically - and ignored when review time comes.

So I built two prompts that use it. One for managers, one for employees.

You populate both prompts with your review questions - about 10 minutes for the whole company. Then send them out with your review cycle announcement, with instructions like this:

“Dear team, paste this prompt into your Claude Cowork, Granola, or another notetaker - and review the mid-year draft it gives you. Polish it, then paste the final version when submitting your review. If you have context from meetings with your people, connected docs, or Slack (connect them) - the review will take 15 minutes instead of being a document tax.”

That’s your whole rollout. No new tool to buy, no training session to run.

They just put in the prompt and get structured results - ready-to-use answers to your actual review questions, built from six months of their own context. Every claim tied to a specific example. Where evidence is thin, it says so instead of inventing.

They read, adjust, submit. That’s it.

Writing time drops from about 60 minutes to 15. Across 50 managers and 200 employees, that’s weeks of work you just gave back to the org.

One thing matters here: this works best in Claude Cowork with connections or a notetaker that holds their meeting history - not a plain AI chat. A regular chat only knows what you paste into it. Cowork with connections finds the evidence itself.

Of course managers don’t want their 1-1 notes readable by anyone in the org. Employees don’t want their private conversations feeding some central system.

In this approach, nothing leaves the individual’s environment. Each person runs Cowork with their own connections - the manager’s notes stay in the manager’s notetaker, the employee’s context stays in their own tools. No shared workspace, no org-wide AI digging through everyone’s meetings.

Each person generates the draft privately, reads it, edits it - and then copy-pastes into the company review form. 100% control over what goes in. The raw notes never travel anywhere.

That’s the part that makes this actually deployable. You’re not asking anyone to give up their private context. You’re giving them a faster way to use it themselves.

Both prompts are free.

Send me a connection request and comment MIDYEAR on my LinkedIn post - I’ll send them over.

Comment MIDYEAR and I'll send them over