Job description management software that levels every sentence against your job architecture and puts the legal pay range on every posting
Every job levelled against its standard, and every posting with its legal pay range.
See one case, screen by screen ↓Job Description Management is job description management software that drafts, standardises and levels job descriptions against your job architecture, for HR and compensation teams. Seven agents write the description in your template, level every sentence against your level standards, argue requirements with evidence and put the legal pay range on every posting. The compensation lead approves the level, the hiring manager approves the text, and only then is it published.
How to choose AI solutions for back-office and enterprise work →
Updated
A large employer carries thousands of job titles, and a description for most of them, written over twenty years by hundreds of managers in a dozen formats. A hiring manager who needs a new role copies the nearest old description and inflates it: “leads company-wide data strategy” for a senior analyst, “10+ years and a master’s” for a job its own people do with five. The compensation team can only defend the level by reading every line against the level standards, and the argument about years of experience comes back every month.
Then the recruiter has to find the right range for every place the job is posted. Colorado, Washington, California, New York, Illinois and a growing list of other places require the pay range in the posting, several add a benefits summary, Colorado adds an application deadline, and the EU Pay Transparency Directive adds gender-neutral titles and no pay-history questions. A range is only defensible when the job is correctly levelled — and a description is evidence of what the job really is.
Estimated split of the demo company’s 19-day median in 2025, by hand, and of the 2-day target with the agents.
Seven agents draft, level every sentence, check each other and price every posting; the compensation lead and the hiring manager decide and sign.
Kestrel & Vane Group, about 40,000 people in 30 countries, needs a Senior Data Analyst – Commercial (JD-2291) in Denver and Chicago. The hiring manager sends eight lines of notes. Here is what happens next, screen by screen, in the working solution.
Priya owns the job architecture: 18 families, 1,148 jobs. Her morning opens on what the agents did overnight — 11 new drafts levelled, 3 with sentences above their level; 34 postings re-checked against 12 jurisdictions, 1 blocked for a Dutch title that is not gender-neutral; 3 duplicate groups covering 2,469 people. Below it, the six level decisions that need her, each with the agents’ reading in one line.
“3 sentences read as Principal; Marcus agreed the rewording. After it, 93%.”
Marcus Delgado, Director of Commercial Analytics, pasted eight lines — ending “Must have 10+ years and a master’s. Rockstar who hits the ground running.” The drafting agent wrote 14 sentences in the group template, kept his requirements as written and dropped “rockstar”, which is on the group’s language list. The levelling agent placed each sentence on the Data & Analytics ladder: 10 at P3 Senior, 3 at P5 Principal, 1 at P2. Fit: P3, 71%.
Drafting agent: “I kept “10+ years” and “master’s” as written: they are Marcus’s requirements.”
Priya clicks the highlighted phrase. The level standard it matches opens beside the text: P5 Principal, “Sets analytics strategy and standards across the company”, from level standards v2026.1, approved by Daniel Okoro. Under the sentence sits the agent’s reason, its proposed wording and Accept, Edit, Reject or Ask why.
Proposed: “Contribute to the Commercial analytics roadmap and apply the group’s analytics standards.” · Marcus agreed in the thread, Wed 16:20.
Marcus accepts the strategy and Executive Committee rewording but pushes back: “I want 10 years — this person runs pricing on their own.” The language and requirements agent answers that running pricing alone is P3 autonomy, “chooses methods without direction”, and that four of the seven people in Senior Data Analyst roles have fewer than 10 years. The peer agent had already compared the job with the approved P3 jobs on the rung and agreed with P3. Marcus: “OK, 5+.”
Peer evidence is shown only as an aggregate of 5 or more people; no one’s record is shown.
The pay transparency agent priced the job from the 2026 pay structure for P3, Data & Analytics: Denver $96,000–$124,000 and Chicago $99,000–$128,000, with the benefits summary and an application deadline of Fri 13 Nov 2026. Colorado’s rule — range, benefits, deadline, career progression notice — and Illinois’s — range and benefits — are met. Adding Seattle prices it at $104,000–$134,000 and checks Washington’s rule before anyone can publish.
The pay transparency agent never invents or widens a range; an exception goes to the Head of Total Rewards.
One click applies the 5 changes Marcus agreed in the thread. The dots on the ladder move to P3 and the levelling agent re-reads the job: every sentence at P3 Senior, 93%; ranges unchanged. Priya approves level P3 with reason code L1 · Duties match the level standard and her e-signature. Had she approved with sentences still above the level, only L4 · Override, with a note, would do.
“I approve this as Priya Shah, Compensation Lead — e-signature, Thu 15 Oct 2026, 09:14.”
The text goes to Marcus once the level is set. He sees exactly what candidates will read — version 4, 13 sentences, level P3, two locations with their ranges — and approves with T3 · Approved; requirements agreed in the thread, and his own e-signature.
Publishing stays disabled until the level and the text are both approved. Owen publishes; the publishing agent sends version 4 to the recruiting system and career site and confirms in the thread: Denver $96,000–$124,000 and Chicago $99,000–$128,000, closing Fri 13 Nov, REQ-40817 confirmed and the HR system holds the job at P3.
Publishing agent: “Published v4 with both approvals.”
The audit trail lists every action by people and agents, hash-chained: Marcus’s request with 8 note lines, the draft, the levelling, the peer check, the ranges, the 5 proposals he agreed, Priya’s level approval with L1, Marcus’s text approval with T3 and the postings. Nothing in it can be edited or deleted, and one click exports all the evidence for the job.
Daniel Okoro, Head of Total Rewards, has opened a change to the P3 standard: typical experience from 4–7 to 3–6 years. The architecture steward re-reads the library first: 9 jobs re-read, 0 change level, 14 requirement lines to reword, 61 people in those jobs and 2 live postings to re-check. A standard change needs two signatures and this preview first.
“No job changes level: duties decide the level and experience is a guide.”
The same job, seen by the four people who carry it — what their month looked like, and what it looks like now.
Seven agents draft, level every sentence, check each other and price every posting; the compensation lead and the hiring manager decide and sign.
Writes or rewrites a description from the hiring manager’s notes, an old description or a nearby job, in the group template, and says what it kept, reworded and dropped.
Levels each sentence against the family’s level standards — scope, autonomy, impact, people and experience — cites the standard line it matches, and proposes the fit and the wording for outliers.
Checks the levelling against the approved jobs on the same rung and the experience and education of the people already in the job, and says whether it agrees.
Flags coded words, requirements the level does not need, degree inflation, titles that are not gender-neutral and pay-history questions, proposes plain alternatives and argues them in the thread with the evidence.
Prices the job from the pay structure for each posting location, adds the benefits and closing date, checks every jurisdiction’s posting rule, and re-checks on every change.
Sends the approved version to the recruiting system and career site and the job to the HR system, and confirms back in the thread.
Reads the whole library every week for duplicate jobs, stale descriptions and levels that have drifted, and previews the impact of a level-standard change before it is signed.
Approve level and text. The agents propose; a named person decides.
Ask about any job, level, range or posting — or tell the agents what to do. Every answer cites its source, and nothing changes until a person approves it.
Why not 10 years?
The P3 standard is typically 4–7 years. Of the 7 people doing this job, 4 have fewer than 10 years — a 10-year rule would exclude most of the people who already do it well. “5+ years, or equivalent” keeps the bar at the level.
Why P3 and not P4?
P4 leads a domain across several businesses in a division. Every duty here is for the Distribution business alone, and the peer agent found the same scope in the three approved P3 jobs. If Marcus adds KV Parts Online pricing, I would read it as P4.
What changes if we also post in Seattle?
Seattle is Zone A: the P3 range is $104,000–$134,000. Washington needs the range and a benefits description in the posting; both are ready. I have added Seattle to the postings.
Why is the Rotterdam posting held back?
Its Dutch title “Verkoopingenieur” reads as masculine, and the directive requires gender-neutral vacancy notices and titles. The range €62,000–€78,000 is already in it. I propose “Sales Engineer (m/v/x)”; Owen decides.
13 screens from the working solution, on its sample data. Pick one to see it large.
The thresholds the agents work to and the approval clocks are settings, saved as a new version and logged.
| Setting | Default | Choose from |
|---|---|---|
| Lowest fit the agents may propose without a person | 80% | Any percentage; below it the levelling agent asks instead of proposing |
| Flag a sentence when it reads this many levels away | 1 level | Number of levels |
| A description is stale after | 24 months | Number of months; then the owner gets a review task |
| Largest range spread allowed | 30% | Maximum ÷ minimum − 1, as a percentage |
| Peer evidence only from groups of at least | 5 people | Number of people; never an individual’s record |
| Application window when a law requires a closing date | 30 days | Number of days |
| Time for the level decision | 2 business days | 1 · 2 · 3 · 5 business days |
| Reminder to the approver | At 75% of the time allowed | At 75% · at 50% of the time allowed |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (Harvard Business School and Burning Glass, Dismissed by Degrees, 2017 — 6.2 million US jobs at risk of degree inflation; Colorado Equal Pay for Equal Work Act, C.R.S. 8-5-201 — $500–$10,000 per violation). Directive (EU) 2023/970 Art. 5, California Labor Code §432.3 and Washington RCW 49.58.110 are named as context. 19 days is the demo company’s own figure. Kestrel & Vane Group, its jobs and the people named on this page are fictional — characters and sample data in the working solution.
Job Description Management is job description management software that drafts, standardises and levels job descriptions against your job architecture, for HR and compensation teams. Seven agents write the description in your template, level every sentence against your level standards, argue requirements with evidence and put the legal pay range on every posting. The compensation lead approves the level, the hiring manager approves the text, and only then is it published.
The levelling agent places every sentence on the job family’s ladder and cites the level-standard line it matches — scope, autonomy, impact, people or experience. A description that reaches above its level shows at a glance. A peer agent checks the result against the approved jobs on the same rung and the people already in the job, and both views go to the compensation lead.
Yes. The pay transparency agent prices each posting location from your pay structure, adds the benefits and closing date, and checks the location’s posting rule — the working solution carries rules for 12 jurisdictions, including Colorado, Washington, California, New York, Illinois, Ontario and EU member states. It re-checks every posting on every change and never invents or widens a range.
It checks EU postings for what Article 5 of Directive (EU) 2023/970 asks: the initial pay or its range, a gender-neutral title and no pay-history question. Every level decision cites the standard line behind it, so the reasoning for each job’s level is on the record. Until employment counsel confirms a country’s national law, the agents apply Article 5 as written.
No. The peer agent uses only aggregates of 5 or more people — for example, the median experience of the people in a role and how many have a master’s degree. No one’s individual record is shown.
Yes. The job description template, the level standards, the language list, the requirement rules, the thresholds and the approval times are all settings, saved as a new version and logged. A change to a level standard needs two signatures and an impact preview first.
Yes. The agents draft, level, check and propose; they never set a level, change a range, remove a requirement, merge or archive a job, or publish without both approvals. The compensation lead and the hiring manager each sign with a reason code and an e-signature, and every step is on a hash-chained audit trail.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your job architecture, level standards, pay structure, template and a sample of your descriptions — and it goes live once every quality gate has passed. A first-time setup imports your existing titles and has the agents propose the architecture for you to approve.
Every HR answer cited to the policy that applies; sensitive cases always with a person.
1-min demoAI Resume ScreeningEvery applicant read against the job, and every decision cited and checked for fairness.
1-min demoWorkplace Investigation ManagementEvery finding tied to quoted evidence and tested for gaps before anyone signs.
1-min demoWe’ll run Job Description Management on one of your job families, its level standards and a sample of your descriptions.