Every assessment, every interview, every technical rating, every CV screening — they all flow into one place: the assessee's competency profile. Understanding how the inflows are weighted is the difference between using the profile as evidence and using it as a vague impression.
The profile is the spine
The competency profile is a per-person, per-competency aggregate of every measurement made of that person. It is not stored as a static record; it is recomputed on every relevant inflow. When a new assessment lands, the profile is updated immediately for every competency the assessment touched.
The Grow surfaces — gap dashboard, career paths, succession, mobility, learning recommendations — all read the profile. It is the single source of truth for what someone can do, and how confident we are in that.
What contributes
In v0.5, the inflows are:
- CV-derived competency extraction at application time. Lowest fidelity (0.4) — text mining of a CV is weak signal.
- Behavioural assessment results for behaviour-relevant competencies. Mid fidelity (0.5–0.6).
- Cognitive assessment results for cognition-relevant competencies. Mid fidelity (0.5–0.6).
- Technical assessment — self-rating only is fidelity 0.4; self-and-manager rating is fidelity 0.8.
- Structured interview ratings roll up per competency at fidelity 0.7 — both hiring interviews and development structured interviews.
- Simulation results at the highest fidelity in the v0.5 set, 0.9.
The blend
For each competency, the profile holds a single current level value with a confidence score. New inflows update both:
- The level is a weighted average of all contributing measurements, weighted by fidelity and decayed by recency. A simulation result this month carries more weight than a structured interview from a year ago, even though the structured interview's fidelity is also high.
- The confidence rises when independent measurements agree, falls when they diverge. Two independent assessments rating a competency at level 4 produce higher confidence than two assessments where one says 4 and the other says 2.
The full mechanics, including the decay function and the divergence handling, are in Fidelity and recency.
What does not contribute
- Free-form notes a manager writes outside an assessment context.
- Performance reviews from outside Jadiroon.
- Project completion records or output volume.
- Self-reports on the employee's profile page.
The profile is built from observed measurements only. This is a deliberate constraint — it means the profile is auditable (you can trace every level to a measurement) and resists pollution from anecdote.
What the assessee sees
The same profile their manager and HR see, shown on /me/competencies. Each competency has a bar showing the current level, an optional target marker (drawn from the growth plan's target role), and a small badge if the competency is below target.
Click a competency and an evidence drawer opens with every measurement that contributed: assessment type, date, level delta. The assessee can therefore see why their profile says what it says, not just what it says.
Bootstrapping a first profile with AI
A brand-new employee often has no assessments yet, so their profile starts empty against the competencies their position requires. On the person's Competencies tab, an HR admin can click Bootstrap competency profile with AI to draft a starting profile. The model scores only the position's required competencies, grounded in whatever evidence is on file (pasted CV text, CV-extracted competencies, any existing measured levels) and assigns a conservative baseline where there is no evidence — it never guesses competence upward.
The result is a draft the admin reviews in a dialog: each competency shows the suggested level, a one-line rationale, and an editable value. Competencies that already carry a measured score are flagged and left unchecked by default, so an existing profile is never overwritten unless the admin explicitly opts in. Nothing is written until the admin confirms; on confirm the chosen levels are saved through the same authoritative admin-override path as a manual score edit, with a history entry marking them as an AI-bootstrapped draft.
When the profile feels wrong
Two common cases:
- A recent strong performance is not yet reflected. This usually means no recent assessment has been taken in the relevant competency area. Solution: assign a fresh assessment.
- An old weak performance is dragging the profile down. The recency decay handles most of this automatically — old measurements lose weight over time. If a specific old measurement is materially wrong, it can be retired from the assessment record (covered in the report page).
Re-rating the profile manually is not supported. The profile reflects measurements; you cannot edit the level directly.
Next
Section V — Growing your people opens with the profile in plainer language and walks through every Grow surface that reads from it.