AiOS→HR / People→Professional Services→Develop & Grow→HR11
Performance Goals & Review Narrative
Frame performance goals, compare evidence to expectations, and draft a review narrative with a capability-gap summary.
Linking goals to specific evidence gives managers a defensible review narrative and surfaces capability gaps early enough for targeted development.
GenAI Impact
65%
Faster
23.9
Hours saved
36.9
Hours without AI
Based on: 6 direct reports in a single review cycle
Enforces a structured evidence-to-rating-to-narrative chain where every attainment rating and review statement traces to a specific documented observation, eliminating unsupported impressions and ensuring balanced coverage of strengths and development areas across all goal areas.
Prevents personal-disclosure leakage by mandating redaction of non-work self-assessment content before prompting, and blocks hallucinated evidence attribution through mandatory verification of every AI-cited observation against the original performance evidence log.
Before You Start
This workflow processes individual performance records, manager observations, and self-assessment reflections. Do not paste these inputs into public or unapproved GenAI tools.
GenAI may misattribute evidence across goal areas or fabricate behavioural examples. Verify every cited observation against the original evidence log before sharing the review.
Who's Involved
Manager
Frames performance goals, verifies evidence accuracy, and approves the final review narrative.
HR Business Partner
Reviews narratives for consistency, checks for bias, and ensures alignment with review cycle standards.
Execution Steps
Before you start
Inputs
Prompt
Structure role expectations into reviewable performance goals
CONTEXT You will be provided with the following source documents: 1. 30-60-90 Onboarding Plan 2. Role Expectations Document 3. Performance Evidence Log 4. Self-Assessment Notes TASK Extract and structure the role's performance goals into a Structured Goals Framework. Each goal must have a clear expected outcome and a measurable or observable indicator so evidence can be mapped against it in later steps. OUTPUT FORMAT Produce a markdown table with the following columns: - **Goal ID** — sequential identifier (G1, G2, G3...) - **Goal Area** — category (e.g., Technical Delivery, Stakeholder Engagement, Professional Growth) - **Goal Statement** — one sentence describing what success looks like - **Expected Outcome** — the observable result or deliverable - **Measurement Approach** — how attainment would be assessed (metric, observation, artefact) - **Time Horizon** — the period the goal covers (e.g., Q1, first 90 days, full year) Include one row per distinct goal. Combine overlapping expectations into a single goal with a clear statement. CONSTRAINTS Do not invent goals not present in the source documents. Do not use organisation-specific jargon — keep goal statements generic and clear. Do not include employee names or identifying details in the output.
Outputs
Verification: Verify that every goal in the framework traces to a specific expectation in the source documents — GenAI may merge or invent goals.
Before you start
Data Handling: Do not include employee personal disclosures from the self-assessment that fall outside work performance — redact before prompting.
Inputs
Prompt
Map documented evidence to each performance goal
CONTEXT You will be provided with a Structured Goals Framework, a performance evidence log containing manager observations, and self-assessment notes from the employee. TASK For each goal in the framework, extract specific evidence entries from the log and self-assessment that relate to that goal. Produce an Evidence Mapping Table linking every goal to its supporting or contradicting evidence. OUTPUT FORMAT Use a markdown table with the following columns: - **Goal ID** — matches the Structured Goals Framework - **Evidence Source** — [Evidence Log / Self-Assessment] - **Evidence Quote** — verbatim excerpt or specific factual observation - **Relevance** — one phrase explaining how the evidence relates to the goal - **Strength** — [Strong / Partial / No Evidence] Include one row per evidence entry. If a goal has no matching evidence from either source, add a single row with "No evidence found" and strength "No Evidence". CONSTRAINTS Do not infer performance outcomes not stated in the source materials. Use verbatim quotes where possible. Do not include personally identifiable information beyond what is necessary to attribute the evidence.
Outputs
Verification: Verify that each evidence quote exists in the original log or self-assessment — GenAI may paraphrase or fabricate observations.
Before you start
Inputs
Prompt
Outputs
Verification: Verify that attainment ratings are consistent with the evidence counts — GenAI may inflate ratings for goals with sparse evidence.
Inputs
Prompt
Outputs
Verification: Verify that evidence citations in the narratives match actual entries in the Evidence Mapping Table — GenAI may fabricate supporting examples.
Inputs
Prompt
Outputs
Verification: Verify that every listed gap traces to a Partially Meets or Below rating — GenAI may surface spurious gaps from fully met goals.
Before you start
Inputs
Prompt
Outputs
Reference
Guardrails
- Evidence-Linked Statements Only — Every claim in the review must cite a specific documented observation or output — no inferred or assumed performance.
- Separate Evidence From Judgment — Complete evidence extraction and mapping before rating attainment or drafting narratives to reduce confirmation bias.
- Balanced Narrative Structure — Each goal narrative must include at least one strength and one development area — avoid deficit-only or praise-only reviews.
Pitfalls
- Pasting raw self-assessment content containing personal disclosures into the GenAI prompt without redaction.
- Accepting AI-generated evidence citations without verifying them against the original performance evidence log.
- Allowing the AI to produce positive-only narratives that omit development areas for highly rated goals.
- Using AI-generated attainment ratings directly without checking they match the evidence strength distribution.
- Including compensation or promotion language in the review narrative prompt context, contaminating the output.
Definition of Done
- Every review narrative cites at least one specific evidence entry per goal area from the original evidence log.
- The capability gap summary traces each listed gap to a goal rated Partially Meets or Below in the attainment analysis.
- The performance review pack contains all three sections — narratives, gap summary, and attainment overview — in a consistent structure.
- No narrative contains vague praise or deficit-only language — each balances strengths and development areas.
Unlock the Full Library
Get full access to all prompts, execution steps, and downloadable examples — for this playbook and the rest of our GenAI capability framework — AGASI AiOS.
We'll send a magic link — no password needed.
AGASI AiOS · HR11 v1.0 · Apr 8, 2026