Self-assessment season,with receipts
Every review cycle, the same scene: a blank form asking what you accomplished, and a memory that reaches back about three weeks. If you work through Khint, the answer already exists. Your work sessions hold the tickets you shaped, the pages you wrote, the decisions you logged. Copy the recaps, run one AI action, and the first draft is evidence, not guesswork.
The review asks for six months,
your memory offers three weeks
The unfairness of self-assessments is structural: the people doing the most fragmented, glue-style work, POs and BAs above all, are the ones whose impact is hardest to reconstruct. Nobody remembers the ambiguity you removed in March. Khint doesn't make you a better archivist; it makes archiving a side effect. Agents run your drafting, Capture pulls evidence off the screen, and Memory quietly keeps the record of both, so when the form arrives, the raw material is already written.
From logged quarter to submitted review
Six steps, most of which happened while you were just doing your job.
During the quarter: work with sessions on
No extra habit needed, just name a Memorysession per initiative, “Checkout revamp”, “Q3 discovery”, and work normally. Every AI action you run, every capture, every Jira ticket or Confluence page you file from the palette lands on the active session as a timestamped entry. Your evidence collects itself while you do the job.
Review season: copy each recap
The self-assessment form arrives. Instead of staring at it, open Memory and hit Copy recapon each session that mattered this cycle. Each click puts the session's compacted summary plus its entry list on your clipboard as clean markdown. Paste them all into one working document: that's your quarter, in writing, in minutes.
Run one Action: draft the self-assessment
Save an Actionfor this once. In the editor, describe it in a line, “turn work recaps into an honest self-assessment that maps each item to its impact”, and press Generate to let Khint draft the prompt. Then select your pasted recaps, hit Cmd+Shift+K, run it, and a structured first draft replaces the blank page: what you did, why it mattered, grounded in the entries.
Answer the form, question by question
Company forms ask specific things: “biggest impact”, “what would you do differently”. Because Actions run on whatever text you select, you can work per section: select the relevant recap material, run a follow-up Action shaped for that question, paste the answer into the form, move on. The same evidence feeds every box.
Polish the tone before you submit
Self-assessments fail in two directions, chest-beating or self-erasure. A final rewrite Action, “tighten this, keep claims factual, cut filler”, gets the register right. If English isn't your first language, the same pass fixes that too. Polishing English with AI →
Keep the sensitive parts on your machine
Review material is personal by definition. Your sessions never leave your machine, Memory is local-only, and an Action only ever sees the text you selected for that run. For passages that name colleagues, switch on the Action's redact-PII option. How PII redaction works →
Why the draft rings true
AI-written self-assessments have a reputation problem: managers can smell a review generated from nothing, all confident verbs and no specifics. The fix isn't a better prompt, it's better input. When the Action runs on your session recaps, the specifics are supplied, not imagined: the actual ticket, the actual page, the actual week it happened. You still edit, you still own the claims, but you're editing a draft grounded in your own record. That's also the honest division of labor: Khint remembers, the AI structures, you judge. How the work journal builds itself →
Common questions
I didn't keep sessions running this quarter. Is this useless to me now?
No, it just works in two speeds. For the review that's due now, gather whatever raw material you have, shipped tickets, status emails, sprint review notes, paste it into one document, select it, and run the same drafting Action: it works on any text you select. Then start a named session for each initiative going forward, and next cycle the evidence collects itself while you work.
Does my work log or my self-assessment get uploaded somewhere?
Your sessions don't: Memory is local-only, sessions and their entries live in a SQLite database on your machine and are never synced to Khint's servers. When you run an AI Action, only the text you selected for that call (plus the session summary, if you opted the Action into it) is sent for processing. And for review material that names colleagues, you can switch on the per-Action redact-PII option, which replaces emails and phone numbers with placeholders before the text is sent to the AI.
How is this different from asking ChatGPT to write my review?
The input. A chatbot starts from what you can remember about six months of work, which by review season is mostly the last three weeks. Khint starts from the record: Copy recap hands you each session's summary and its actual entries, the tickets filed, the pages written, the decisions noted, so the draft cites things that happened instead of generic accomplishment-speak. Mechanically it's also lighter: select text, hit the shortcut, paste the result, no tab juggling.
Can I use the same flow for a promotion case or a brag document?
Yes, it's the same material with a different prompt. Save a second Action, one that argues scope and impact for a promotion packet instead of answering a review form, and run it on the same recaps. Some people also paste a recap into a running brag document at the end of every month, so review season starts from a document that already exists.
Next review season starts today
Free with 10 AI actions and 5 captures per day. No credit card. Start a session now, and the next form meets a full record instead of a blank memory.