The Review Agent in Workleap Performance helps you prepare and write a performance review in one conversation — it gathers the evidence that matters, then drafts in your own voice, while you keep every judgment and the final word.
What the Review Agent does
The agent works with you inside the review form, in two steps you move between freely:
Prepare — the agent gathers and organizes the evidence for the person you're reviewing, names patterns, and points out gaps, so you understand the period before you write.
Write — the agent drafts answers in your voice, one suggestion at a time, and you refine, shorten, or rewrite anything.
The review form stays on the right the whole time. A Review context panel keeps every piece of evidence the agent surfaces, and it grows as you talk to the agent.
Workleap AI insight — The agent applies Workleap's performance-management expertise by default: it organizes the facts, surfaces patterns and contradictions, and keeps answers specific and balanced. The synthesis is the agent's job; the judgment stays yours.
Who can use the Review Agent
The Review Agent is available to anyone completing a review in a review cycle — no special role needed. If you've been assigned a review, you can use the agent to prepare and write it. What the agent can access depends on the review type you're filling out (see What the agent can access for each review type below).
The agent is available across all four review types:
Self-review — you, about your own work.
Manager review — a manager, about a direct report.
Peer review — a peer, about a colleague they were asked to review.
Upward review — a direct report, about their manager.
The agent never changes what you're allowed to see. It works strictly within your existing permissions and the review type you're completing.
Access your review
You can open a review in two ways.
From Workleap Performance
Log in to Workleap and go to Performance.
Open the Reviews section.
Select an active review.
From a notification
Select Start review in the email or notification you received about the review.
How the agent prepares your review
Log in to Workleap and go to Performance.
Open the review you need to write.
In the conversation panel, select Help me prepare the review (or ask the agent in your own words).
Review what the agent surfaces in the Review context panel, and tell it what to dig into.
The agent is steerable: it asks before pulling additional sources, and you decide what enters the context. When a source has nothing for the period, the agent says so rather than inventing content.
Tip — You stay in charge of the preparation. Ask the agent to focus on a specific goal, project, or question, and it pulls only what's relevant to that.
What data the agent can use
The agent draws on two kinds of sources, and it only ever uses data the reviewer already has access to.
Workleap sources (gathered when you start preparing):
The current review cycle (its period and questions) and the previous cycle
Current and past reviews for the person
Goals the person owns or contributes to during the cycle
1-on-1 agendas
Recognition (Good Vibes) received and sent
Connected apps (your personal productivity apps — the agent uses only the ones you've connected):
Jira — your work items and projects
Notion — pages and databases in your Notion workspace
Slack — channel and direct messages you have access to
Microsoft Outlook — your Outlook emails and calendar events
Microsoft Teams — messages and channels in your Teams workspace
Google — your Google files, emails, and calendar events
Confluence — knowledge from Confluence spaces and pages
When the person being reviewed manages a team, the agent can also offer team-level context such as engagement scores and feedback — proposed, never pulled automatically.
Important — Workleap AI only uses sources the reviewer already has access to. It never exposes data you couldn't otherwise see.
Connect your apps
An Administrator first allows personal app connections for the organization (the Users can connect to personal integrations policy).
Once allowed, log in to Workleap and open your profile, then go to Connected apps.
Select Connect next to each app you want the agent to use.
Each person connects their own apps, and the agent pulls only from the apps that person has connected.
Add files
You can upload files to give the agent extra context for a review.
Up to 10 files at a time (no overall limit per conversation).
Up to 32 MB per file.
Supported file types:
Documents — PDF (.pdf), Word (.doc, .docx), OpenDocument Text (.odt), Rich Text (.rtf), plain text (.txt), Markdown (.md)
Spreadsheets — Excel (.xls, .xlsx), CSV (.csv), tab-separated values (.tsv)
Presentations — PowerPoint (.ppt, .pptx)
Images — JPEG (.jpg, .jpeg), PNG (.png), BMP (.bmp), TIFF (.tif, .tiff), HEIC (.heic)
Emails — standard email (.eml), Outlook email (.msg)
Web and other — web page (.html, .htm), EPUB (.epub), XML (.xml), reStructuredText (.rst), Org-mode (.org), digital signature file (.p7s)
What the agent can access for each review type
The agent only ever sees what the reviewer is allowed to see, and that scope is set by the review type. This protects everyone's privacy and keeps each review grounded in the right evidence.
Manager review (manager about a direct report) — Full access to the person's stack: reviews, goals, 1-on-1s, recognition, and connected apps.
Self-review (employee about themselves) — The full stack, scoped to the reviewer's own data.
Peer review (peer about a peer) — Shared work only: goals both people contribute to, recognition exchanged between them, and public or co-owned work in connected apps. No access to 1-on-1s or private history.
Upward review (report about their manager) — Limited to the reviewer's own experience of the manager: recognition exchanged between them and the manager's work the reviewer can already see. No access to the manager's other reports' data.
Important — Some "no data" is an access boundary, not an absence. For example, a peer reviewer doesn't see 1-on-1 history by design — so the agent won't surface it. The agent explains the boundary instead of leaving you guessing.
How the agent helps you write
When you're ready to write, the agent asks how you want to work and adapts to your choice. You can switch modes at any time, question by question.
Draft it for me — The agent drafts every answer in your voice, then steps back. It tells you which answers are still thin on evidence so you can decide what to do next.
Walk me through it — The agent leads one question at a time, offering a single suggestion you accept, edit, or ask it to rewrite.
I'll write it myself — The agent stays out of the way and helps only when you ask.
The agent matches your voice from your past reviews. It offers one suggestion at a time — never competing versions — and never overwrites the edits you make.
Workleap AI insight — When evidence is thin, the agent writes a short, honest answer and flags it, rather than padding it with generic content. An honest gap is more useful than a confident guess.
What the agent won't do
It doesn't propose a rating by default. It can lay out what the evidence supports under each plausible rating, but the rating is your decision.
It doesn't overwrite your words. Your edits are the bar; the agent works around them.
It doesn't invent evidence. With nothing to ground an answer, it writes close to nothing and names what's missing.
It doesn't act on its own. It can submit or share the review when you ask, but it never offers to do so on its own — you decide when.
Note — The Review Agent runs on Workleap AI and requires an active Workleap Performance subscription. It assists the writing side only — it doesn't change how reviews are shared with or received by the person being reviewed.
Submit your review
When every required question is answered, select Submit to complete your review. You can reopen and edit it until the review cycle is locked.
You can also ask the agent to submit or share the review for you — it just won't offer to on its own.