AI ToolCompensation

Salary Generation Assistant.

A 10-step wizard that walks a People & Talent partner through a defensible salary recommendation: geographic pre-check, finalist confirmation, role context, experience submission, role-by-role breakdown, total relevant years, comparison to role minimum, anchor assignment, geographic multiplier application, and approval-ready outputs. Illustrative data only — the point is the workflow, not the numbers.

Built in Lovable Structured wizard, not a chatbot Exportable approval email
Try the live demo →
10
Linear steps, each unlocking the next — geographic pre-check through approval-ready outputs
1
Copyable approval email generated at export — ready to send without editing
0
AI "decisions" — a human runs every step; the tool shows the work
TL;DR

Most compensation conversations break down somewhere between "what does the band say?" and "what should we actually offer this person?" The math isn't hard. The governance is — who decided, on what inputs, against which band, with what adjustment, and where is that written down? This demo is a structured wizard that makes the logic visible at every step and produces a copyable approval email at the end. Illustrative data only; your org's bands replace the numbers.

Why I built this

The math isn't hard. The governance is.

Most compensation conversations break down in the same place: somewhere between "what does the band say?" and "what should we actually offer this person?" Who decided, on what inputs, against which band, with what adjustment — and where is any of that written down?

Salary decisions made under time pressure tend to live in Slack threads and institutional memory. That's how you get inconsistency, failed audits, and employees who eventually compare notes. The risk isn't the math — it's the undocumented process behind the number.

The solution

A structured wizard that shows the work at every step.

This is a demo, not a chatbot. Every step is a governed form with visible inputs, visible math, and a visible audit trail. By the time you reach the export step, every number on screen is traceable to an input two steps earlier.

The bands and multipliers are illustrative — the point is the workflow. Your organization's salary data replaces the numbers; the structure stays the same.

Decision support beats automation. The recommendation is always one click away from being overridden — and the override is visible.

How it works

Ten steps. Each one unlocks the next.

The wizard is linear and gated — you can't skip ahead. Every number on the export screen is traceable to a step you completed earlier.

1
Geographic pre-check
Lock the location multiplier before anything else. Standard, high-cost, or lower-cost region — this multiplier is set once and applied at the end, so it can't be reverse-engineered later.
2
Finalist confirmation
Confirm the candidate is an active finalist before any compensation work begins. Keeps the workflow from running on speculative or pre-offer candidates.
3
Role context
Set the role level and track. Level determines the experience minimum and the anchor range that governs every calculation that follows.
4
Submit experience
Enter the candidate's work history. Job titles, organizations, and dates — the raw input that the next step will parse and classify.
5
Role-by-role breakdown
Each role is tagged as relevant, partial, or excluded with a brief justification. The logic is visible — nothing is silently dropped.
6
Total relevant years
Relevant and partial years are summed. The total is displayed explicitly before any comparison to the role minimum.
7
Compare to role minimum
Total relevant years are measured against the level floor. Years above the minimum are surplus — they drive the anchor movement in the next step.
8
Assign anchor
+1 anchor per five full surplus years, capped at Anchor 8. The recommended anchor is shown alongside the logic that produced it.
9
Geographic multiplier
The multiplier locked in Step 1 is applied to the anchor salary. Pre-multiplier and post-multiplier figures are both shown.
10
Approval-ready outputs
A copyable approval summary covering every input, classification, and calculation — ready to send without additional editing.
What it exports

Every variable an approver needs, in one place.

The export step generates a copyable approval email. Every field traces back to a step you completed — nothing is inferred or filled in automatically.

Approval email — illustrative structure Illustrative data only
Role & level
Selected in Step 2
Geographic region
Set in Step 1 — multiplier applied throughout
Target total cash
Band × anchor × geo, calculated in Step 3
Relevant experience
Parsed in Step 4, total and breakdown
Surplus years
Years above level minimum, from Step 5
Recommended anchor
Starting anchor + surplus movement, from Step 6
Proposed salary
Anchor salary with geo applied, from Step 7
Compa-ratio
Proposed vs. band midpoint, from Step 8
How I think about tooling

What this demo is deliberately not.

Not a system of record

Resumes and approval documents belong in your HRIS. This tool produces the recommendation; your systems hold the paperwork.

Not a pricing engine

The bands are illustrative. Your organization's salary data replaces the numbers — the workflow structure stays the same.

Not an AI that decides

A human runs every step. The tool shows the work, enforces the sequence, and makes the logic legible — it doesn't generate an offer on its own.

Boring is a feature

Monospace field labels, serif headings, no animations pretending to be intelligence. It looks like a form because it is a form.

Built-in guardrails

Constraints that are part of the UI, not a policy doc nobody reads.

A persistent note appears on every step reminding the user what this tool is and isn't. The following are enforced throughout.

Steps are linear and gated — you cannot skip ahead or revisit a prior step mid-flow.
Anchor movement is capped at Anchor 8 regardless of surplus experience.
Every number on the export screen is traceable to an input completed earlier in the wizard.
The recommendation is always one click away from being overridden — and the override is visible in the output.
This tool supports decision-making. It is not a system of record. Resumes and approval documents belong in approved HR systems only.
Who it's for

Comp partners, People Ops leads, and founders tired of salary decisions living in Slack threads.

Compensation decisions compound. An underdocumented offer creates problems long after the hire — equity reviews surface inconsistencies, audits expose missing rationale, employees compare notes. The cost of a poorly documented placement decision carries forward into performance conversations, equity reviews, and the employee's understanding of their own standing.

This demo is for anyone who wants to see what a lightweight, governed alternative looks like before commissioning the real thing. The workflow is the deliverable — the illustrative numbers are just there to make it feel real.

← Back to AI tools