01
There Was No Way to Prepare
No tool put goal setting, likely outcomes, and how to ask for them in one place. Everyone worked it out alone, from scratch, every cycle.
What it cost them
A career conversation walked into unprepared
Client
Fairness Factor
Industry
HR Tech & Employee Advocacy
Surfaces
Web App · Admin Console
Scope
Agentic AI · Product engineering
Case Study · HR Tech · Performance Review Preparation
Your Review Is In Three Weeks. You Have No Idea What to Say.
Kodexo Labs built Fairness Factor, a platform where an employee talks through what they want out of their performance review, gets a score for how likely they are to get it, and walks out with the legal brief, talking points, and negotiation playbook to go and ask for it.
4 hrs
of prep reclaimed per review cycle
$150
per user avoided in legal consultation
6
prep documents generated in one flow
01 · Overview
A Coach in the Weeks Before the Conversation That Sets Your Pay.
An employee signs in and starts talking to Sidekick, the platform's advocacy agent. It asks the mandatory things first, position, review date, company, current pay, then asks what they actually want out of the review. A raise. A title. A different manager. Better hours.
For each goal the agent works through the factors that decide whether it lands and scores it. By the end the employee holds a Goal Attainability Report, six prepared documents, a rehearsal against an AI playing their manager, and drafted emails to send. Sessions pause and resume, and the Vault keeps every document and transcript.
Up to 7 goals per user6 prep documents per sessionPOC shipped, v1.1 specified
02 · The Problem
Everyone Else in the Room Prepared. The Employee Improvised.
A performance review decides pay, title, and the next two years of someone's career. The manager arrives with a form, a rubric, and HR behind them. The employee arrives with a hope. The specification names four problems.
02
Nobody Could Say What Was Realistic
Ask for 5% or 15%? Push for the title now or wait? Nothing turned the facts of the job, the policy, and past cases into a number.
Why it hurt
Career-defining asks made on a hunch
03
The Prep Material Was All Generic
Rights, pay benchmarks, talking points, negotiation advice: pieced together from articles that knew nothing about their job, their company, or their manager.
Where it went wrong
Hours of research that fit nobody's situation
04
There Was Nowhere to Practice
The first time anyone said the words out loud was in the real meeting, with the real manager, and the real answer on the line.
What was missing
One rehearsal, and it counted
Swipe →
01
There Was No Way to Prepare
No tool put goal setting, likely outcomes, and how to ask for them in one place. Everyone worked it out alone, from scratch, every cycle.
What it cost them
A career conversation walked into unprepared
02
Nobody Could Say What Was Realistic
Ask for 5% or 15%? Push for the title now or wait? Nothing turned the facts of the job, the policy, and past cases into a number.
Why it hurt
Career-defining asks made on a hunch
03
The Prep Material Was All Generic
Rights, pay benchmarks, talking points, negotiation advice: pieced together from articles that knew nothing about their job, their company, or their manager.
Where it went wrong
Hours of research that fit nobody's situation
04
There Was Nowhere to Practice
The first time anyone said the words out loud was in the real meeting, with the real manager, and the real answer on the line.
What was missing
One rehearsal, and it counted
03 · The Solution
One Conversation, Six Stages, Nothing Lost in Between
The whole product is a chat on the left and whatever the chat is producing on the right. Pick a stage to see the screen the employee is actually looking at.
Stage 01
Pick Up Exactly Where the Last Session Stopped
Prep never happens in one sitting. Sidekick greets the user by name, says which stage they left mid-way, and keeps every past session in the sidebar, ready to reopen.
Resume mid-stage
Google or email sign-in
New goal any time
Stage 02
Say What You Actually Want, in Your Own Words
The employee describes the outcome they are after. Known goals load an expert question set; unfamiliar ones are defined together, then scored with a general set and a stated lower confidence.
50 to 500 characters
Known and new goals
Stage rail navigation
Stage 03
Every Answer Moves the Score
Sidekick asks the factors behind the goal one at a time, or hands over the whole set as a form. Add a factor of your own, weight it, and the estimate recalculates in front of you.
Core vs Enhanced score
User-weighted factors
Highlight text to ask
Stage 04
The Stage That Makes Everything Else Personal
Achievements, KPIs, hard situations, strengths and gaps, plus how the manager communicates and who signs off. Uploads happen in the chat: the AI reads the file and pulls the answers out of it.
Four context sections
PDF or DOCX, 10 MB
Feeds every document
Stage 05
Your Employer's Own Paperwork, Working for You
Policies, KPI documents, the position description, the training manual. Once loaded, the platform argues from the employer's own words instead of generic advice, and keeps them for later sessions.
Search your own files
Clear upload limits
Reused across sessions
Stage 06
Nothing About Your Review Sits in Someone's Inbox
Transcripts, uploaded files, generated reports, and a change log for every document edit, grouped by day. The log records what changed, when, and where in the conversation it was agreed.
Full session history
Document change log
Downloadable prep guide
Stage 01
Pick Up Exactly Where the Last Session Stopped
Prep never happens in one sitting. Sidekick greets the user by name, says which stage they left mid-way, and keeps every past session in the sidebar, ready to reopen.
Resume mid-stageGoogle or email sign-inNew goal any time
What You Walk Away With
Six Documents, and a Rehearsal Before the Real Thing
Every document can be changed by talking to it, and every change is logged.
01
Legal Brief
Your rights in your state, in plain words
02
Salary Research
What your role pays in your city
03
Talking Points
What to say, strongest evidence first
04
Negotiation Playbook
Answers for objections and delays
05
Company Future
Questions about where you fit next
06
Self-Assessment
Your employer's form, already filled in
Role-play
Practise on an AI Playing Your Manager
Pick the moment you are dreading, set the tone, and run it as many times as you like.
Emails
The Message, Written Before You Have To
The meeting request, the follow-up, the reply to a no, drafted, editable, ready to copy.
06 · How It Works
From Conversation to Walking in Prepared
No dashboards to learn. One conversation carries you from stating a goal to walking into the review ready.
01
You Talk, Sidekick Listens
Tell it what you want out of your review: a raise, a title, better hours. It asks the follow-up questions that matter.
02
It Scores Your Odds
Your answers, plus anything you upload about your job and manager, turn into a plain likelihood score for that goal.
03
It Builds Your Case
Legal rights, salary data, talking points, and a negotiation plan, written for your situation, not generic advice.
04
You Rehearse Before It Counts
Practice the real conversation against an AI playing your manager, then walk in with everything ready.
Swipe →
The score updates as you add more detail, and nothing about your review sits in someone else's inbox: every document, transcript, and edit stays in your own Vault.
07 · Outcomes
What Changes for the Person Walking Into the Room
Real time saved, real money saved, better odds at the table. This is what a well-built AI product can do for your users, and it can be built for yours too.
4 hrs
Reclaimed per review cycle
Six prep documents generated in one flow instead of assembled by hand from scratch.
$150
Legal cost avoided per user
A state-specific legal brief answers what most people were paying for consultation time to hear.
+12 pts
Scoring confidence from Enhanced Factors
Letting users add their own weighted factors closes most of the gap on goals with no expert set.
4
Sessions per user, per cycle
Up to seven goals does not fit one sitting. The Vault is what makes coming back cheap.
+25%
Higher follow-up survey completion
Reminders start the day after the review date and repeat every three days until answered.
65%
Of sessions produce a document actually used
One-click copy on every draft, and a change log behind each edit the user agreed to.
Swipe →
FAQ
What Founders Ask Before Building This
The questions that come up whenever advice has to be personal, scored, and defensible.
What Happens When the Goal Isn't in the Knowledge Base?
The agent works with the user to define the goal clearly, then uses a generalized factor set to score it. That score is shown with a disclaimer stating it has lower confidence than a goal with expert factors behind it. The system says what it does not know rather than pretending.
Is the Score a Prediction or a Guess?
It is a weighted calculation over factors that human experts defined and weighted, not a number the model invents. When the user adds factors of their own, they assign the weight and the engine recalculates with a dynamic weighted average, which is why both scores are shown side by side.
Can a User Change a Generated Document?
Yes, by discussing it. They highlight a passage, talk it through with the agent, and once they accept a change the document is rewritten. Each edit is written to a Document Log with a timestamp and a link back to the point in the conversation where it was agreed.
What If Someone Leaves Halfway Through?
The stage, the documents, the uploads, and the transcript are stored. On the next sign-in the agent greets them and says where they stopped. The same path covers adding or changing a goal after the goal stage was already finished.
How Are Uploaded Documents Handled?
PDF or DOCX up to 10 MB. The AI extracts answers to the mandatory and factor questions from the file, stores it in the user's Vault, and makes it retrievable for the rest of the session. Oversized or unsupported files are refused with the accepted formats listed.
Why Cap It at Seven Goals?
Because there is a limit to how many outcomes one person can pursue in a single review. The cap keeps the prep focused and the scoring meaningful rather than producing a list nobody can act on.
Have Advice Worth Charging for and No Product to Put It In?
Twenty minutes with the team that shipped this. Bring the expertise you want scored and delivered, and leave with a straight answer on what it takes to build.
51
Products Shipped
Across 25+ industries
94%
Client Retention
Across every engagement
Top
Rated on Clutch
Elite AI Firm designation
Related work
