Promotion Case Drafting Agent
A promotion case drafting agent is an AI agent for performance & talent management that when a manager nominates someone, assembles the evidence against the leveling criteria for the next level (scope, impact, behaviors), drafts the case narrative for the manager to edit, and marks which criteria have thin evidence. For the promotion committee, produces an aggregate view of nomination and proposed-approval rates by org and cohort.
How does the promotion case drafting agent work?
What flows in, what the agent does with it, where a person decides, and what comes out.
Reads from
Leveling framework and next-level criteria · Reviews, feedback and goal history · Level and time in level · Manager's nomination rationale · Aggregate nomination data
AI agent · runs when a manager nominates someone in the promotion window
Promotion Case Drafting Agent
A person decides
Manager nominates; committee decides; Total Rewards sets pay
Produces
Draft case per criterion · Thin-evidence markers · Committee consistency report by org and cohort
What does the promotion case drafting agent do?
When a manager nominates someone, assembles the evidence against the leveling criteria for the next level (scope, impact, behaviors), drafts the case narrative for the manager to edit, and marks which criteria have thin evidence. For the promotion committee, produces an aggregate view of nomination and proposed-approval rates by org and cohort.
What does it produce?
A draft promotion case with evidence mapped to each criterion and gaps marked, plus a committee-level consistency report
Who decides?
The manager decides whether to nominate; the promotion committee decides who is promoted; total rewards sets the pay outcome. The agent drafts and organizes evidence, it produces no readiness score or ranking of nominees.
What systems does the promotion case drafting agent connect to?
Examples of the kind of systems this agent would read from or write to, so you can picture it in your own stack. The actual set is whatever you run.
-
Performance and talent
reviews, feedback and level
-
Leveling documentation
criteria per level
-
Case workflow
nominations and committee packs
-
Analytics
consistency report above minimum group size
What data does it need?
- leveling framework and promotion criteria
- review outcomes, feedback and goal history
- current level and time in level
- manager's nomination rationale
- aggregate nomination data for the consistency report
How would you measure it?
case preparation hours, per cycle; evidence coverage per criterion, per cycle; nomination and approval rates by cohort, per cycle, above minimum group size; promotion appeals, per cycle
What does a first proof look like?
Take last cycle's promotion cases. The agent redrafts a sample from the evidence; the committee chair compares them with the submitted cases and checks the thin-evidence markers.
You'd call it working when
Drafts map evidence to criteria accurately, no nominee is scored or ranked, and the consistency report matches HR's own analysis above minimum group size.
What usually goes wrong?
- Thin-evidence markers can look like a verdict; word them as 'add evidence here'
- Cohort rate reports need minimum group sizes or they identify people
- If nomination requires a drafted case, managers who cannot use the tool are disadvantaged; keep a manual path
What are the guardrails?
- No readiness score, ranking or recommendation of nominees; drafts and organizes evidence only
- Committee report aggregate only, minimum group size enforced
- Manager nomination and committee decision remain the only decision points; agent output is never a decision record
- High-risk under the EU AI Act: documented, human-reviewed, logged; works council consulted where required
- Access to a case limited to the nominating manager, HRBP and committee
What leaves your boundary is set per build; the inputs above are the ceiling, and where the model runs, what it retains, and the DPA are agreed with your security team before anything is connected.
Our read
The job itself is a decision about a named person. This agent exists as decision support only; the decision stays with the person named above.
Some suites offer this as an automated decision. We do not build it that way: the agent prepares, a named person decides, and the record shows who.
a promotion decision about an individual; high-risk under the EU AI Act, and rate disparities by group are legally sensitive
Where it sits in the order
A promotion decision about an individual; only after review and calibration agents have earned trust.
Is a Promotion Case Drafting Agent worth building for your function?
That depends on your numbers, your data, and what else is on the map for you. The strategy month works that out.
Thirty minutes. Bring the number this would move.