AI-Assisted Reviews: Feature Overview

AI-Assisted Reviews is a 15Five Agent that helps managers and employees write higher-quality performance reviews grounded in real data. It uses 15Five's context layer performance history, 1:1 notes, check-ins, goals, peer feedback, and signals from connected work tools to surface evidence and draft review answers so reviewers never start from a blank page.

AI-Assisted Reviews is part of 15Five's connected AI system. It draws from the Context Layer, including data captured by Kona Meeting Assistant. See How 15Five AI works together for how data flows between capabilities.

This article applies to organizations enrolled in the AI-Assisted Reviews beta. If your organization uses the current AI Assist experience, see Complete a manager review with AI Assist.

Who can use the AI-Assisted Reviews agent?

The AI-Assisted Reviews agent is available to review writers when enabled by an administrator.

Review Type Persona Reflect Write Check Discuss
Self Reviews Employee, Manager - -
Manager Reviews Manager
Upward Reviews - - - - -
Peer Reviews - - - - -

Availability

AI-Assisted Reviews Agent is available to all writers of self and manager reviews when enabled by an administrator.

Note: To ensure the highest level of security, 15Five is continually investing in our overall information security program, resources, and expertise. To review 15Five's security documents, and how we handle your company data, see our Trust Website.

How AI-Assisted Reviews works

AI-Assisted Reviews guide reviewers through four phases:

Phase What happens Available for
Reflect Review Moments, sourced evidence organized by theme, before any writing begins Self reviews, Manager reviews
Write AI drafts answers grounded in your Moments. A Review Assistant provides real-time Suggestions and a Chat assistant. Self reviews, Manager reviews
Check Preview how your review reads to the recipient, with a quality summary across multiple dimensions Manager reviews only
Discuss A Conversation Guide helps you prepare for the performance 1:1 with a suggested agenda, talking points, and an IDP draft Manager reviews only

The agent is a thinking partner. It never scores ratings, never blocks submission, and never removes the person writing from the process. Every generated answer can be edited freely, and every claim the AI makes traces back to a specific source.

AI-Assisted Reviews are currently in beta and available only to organizations that have been enabled by 15Five. Features and functionality may change during the beta period.

What review types are supported?

Review type Persona AI-Assisted Reviews available?
Self reviews Employee, Manager Yes (Reflect + Write)
Manager reviews Manager Yes (all four phases)
Upward reviews Employee No
Peer reviews Employee No

What data do AI-Assisted Reviews use?

AI-Assisted Reviews draws from 15Five's context layer, scoped to the performance period set during cycle creation:

  • People and Performance Context — performance history, 1:1 notes, check-ins, objectives, High Fives, Kona insights, competencies, and peer feedback
  • Work Context — signals from connected work system integrations (Slack, Jira, Asana, Google Drive, and others), including project activity, collaboration patterns, and communication signals
  • Added Context — Company Context set by HR, and My Context added by individuals. See "How Added Context works" below for details on who can add context and how it is scoped.

The performance period dates control what the agent draws from. Only data within that date range appears in Moments and drafts.

How Kona Meeting Assistant data is used:

  • When Kona Meeting Assistant is enabled, AI-Assisted Reviews draws from the original meeting transcripts generated by Kona, not from notes that have been edited after the meeting.
  • Both participants in a 1-on-1 can edit Kona meeting notes. Edits are saved to the 1-on-1 agenda but do not change what AI-Assisted Reviews draws from.
  • Kona auto-join can be configured to apply to all 1-on-1s or only to newly created 1-on-1s. Only meetings where Kona is present generate data available to AI-Assisted Reviews.

How Added Context works

Context is scoped by who it is about, not by who wrote it. Every context entry is tied to a specific subject (a person, a team, or the company) and only enters AI outputs about that subject. A manager's note about one of their direct reports is used in outputs about that direct report, such as the manager's Focus Brief for them, Moments for reviews of them, and that direct report's own Focus Brief. It never crosses to a different person.

Context type Who can add it What it feeds
Company Context HR Admins and Company Admins Influences AI outputs company-wide (management philosophy, values, current strategy)
My Context (employee) The individual employee Feeds into that employee's own Focus Brief, their self-review Moments, and their manager's review Moments about them
Review context (manager) The manager writing a manager review Feeds into that specific manager review only. The manager uploads files or adds notes about a direct report to help the AI draft the manager review of that person. This context does not flow into the direct report's self-review.

A manager cannot add My Context on behalf of a direct report. Each person adds their own My Context. Managers add review context per direct report when writing their manager review.

Including historical or pre-Kona notes

If your organization had review conversations before Kona was active and you want that history available to AI-Assisted Reviews, there are two options:

  • My Context (recommended): Upload the historical notes as a file or paste them as text into My Context. This is the most targeted approach because you control exactly what the AI references.
  • 1-on-1 notes: Paste the notes into the shared notes section of a 1-on-1 and end the 1-on-1. The notes will be picked up if they fall within the review cycle's performance period and are relevant to the review questions.

My Context accepts PDFs, Word documents, and text files. You can upload existing documents (strategy one-pagers, development plans, meeting notes from other tools) instead of retyping them.

Support levels

HR Admins choose a support level that controls how much AI guidance reviewers receive. Each level builds on the previous one.

Level Name What it adds
Basic Writing support Writing guidance to structure responses, AI highlights strengths and surfaces observations from context, and AI-generated summaries in calibrations
Medium Manager guidance Everything in Basic, plus manager coaching for review conversations and auto-generated IDPs
Full Full integration Everything in Medium, plus real work data pulled from Slack, Asana, Calendar, and other connected tools

Key differences from the current AI Assist

Dimension Current AI Assist AI-Assisted Reviews (Beta)
Starting point Select data sources, generate a summary Reflect on Moments organized by theme
Draft generation Copy/paste from widget AI drafts all answers directly in the editor
Writing feedback None Real-time Suggestions panel (Actionability, Specificity, Balance, Tone consistency) + Bias detection
Bias detection None Flags gendered language, personality-trait assessments, affinity bias, halo/horn effects, recency bias, and more
Post-writing Submit Check (preview as recipient) + Discuss (Conversation Guide with agenda and IDP)
Self reviews Not available Available (Reflect + Write)

How to request beta access

Contact your account team to opt in to the AI-Assisted Reviews beta. Once enabled, all reviewers in your organization see the new experience during active review cycles.

Organizations not enrolled in the beta continue using the current AI Assist experience with no changes.

Related Articles

Was this article helpful?

Sorry to hear that. Tell us what was missing →