Artificial Idea | AI careers · practical prompts · no hype Thursday, September 25, 2025 · Issue #16 · Prompt Tutorial
The honest audit
How to use AI to write a performance review that actually gets noticed
Most performance reviews are forgettable because they describe what happened. The ones that get noticed make the case for what should happen next.
Fifteen issues. Eight weeks. One consistent argument running underneath all of it: the professionals who navigate the AI transition well are not the ones who panic, and not the ones who ignore it. They are the ones who assess their situation honestly and act on what they find, with specificity and without waiting for institutional permission to do so.
This issue is the tool for doing that assessment.
Before the prompts, one observation about performance reviews specifically, because the two things are more connected than they initially appear. A performance review is not a historical document. It is a strategic one. Its function, when written well, is not to summarise what you did over the past twelve months. It is to make the case for who you are becoming and what you should be trusted with next. The professionals who understand that distinction write reviews that advance their careers. The professionals who do not write reviews that get filed and forgotten.
AI, used correctly, can help you write the former. Here is how.
The self-assessment framework
Before writing a single word of your review, you need raw material that is honest enough to be useful. The most common failure in performance review preparation is starting with the writing rather than the thinking. The prompt below forces the thinking first.
You are an experienced executive coach helping me
prepare for my annual performance review. Your job
is not to make me feel good about the past year.
It is to help me think about it clearly enough
to communicate it strategically.
My role: [job title, level, and brief description
of your primary responsibilities]
My organisation: [industry, size, and any relevant
context about the past year]
Please ask me twelve questions that will help me
surface the most strategically important material
from my past year. The questions should cover:
- Specific outcomes I produced and their measurable impact
- Situations where I operated above my level or beyond
my formal responsibilities
- The most significant challenge I navigated and
what it demonstrated about my capability
- Where I fell short and what I did about it
- How my work connected to the organisation's
most important priorities this year
- What I am capable of now that I was not capable
of twelve months ago
Ask the questions one at a time and wait for my
answers before proceeding. Do not generate the
review until I have answered all twelve.
The instruction to ask questions one at a time and wait for answers is critical. Running this as a dialogue rather than a batch input produces substantially better raw material, because answering one question honestly tends to surface details and connections that would not have emerged from a list-based self-reflection exercise.
Building the narrative
Once you have completed the twelve-question dialogue, use this prompt to turn the raw material into a structured narrative.
You are a senior communications professional helping
me write a performance review that is honest,
specific, and strategically framed.
Here is the raw material from my self-assessment:
[paste your answers from the dialogue above]
My audience: [your direct manager, skip-level manager,
HR, or a combination — be specific about who reads this]
What I want this review to communicate about
my trajectory: [be honest about what you want
to happen next — promotion, expanded scope,
a specific opportunity, recognition of a
transition you have made]
Please write a performance review structured as follows:
1. An opening paragraph that frames the year
in terms of its most significant contribution,
not a list of activities
2. Three to four specific achievement sections,
each built around an outcome rather than
a responsibility, with quantified impact
where the raw material supports it
3. A development section that addresses areas
for growth honestly but frames them as
evidence of self-awareness rather than
as deficits
4. A forward-looking closing paragraph that
connects this year's performance to
a specific next step
Constraints: No generic phrases such as
"I am a team player" or "I go above and beyond."
Every claim must be supported by a specific
example from the raw material.
If the raw material does not support a claim,
do not make the claim.
The constraint against unsupported claims is the one most likely to produce pushback from the model and the one most worth insisting on. Performance reviews filled with generic capability claims are the ones that get filed. Reviews built entirely on specific, evidenced outcomes are the ones that get remembered. The model will attempt to smooth over gaps in your raw material with plausible-sounding generalities. The constraint stops that from happening.
The quantification prompt
One of the most consistent weaknesses in self-written performance reviews is the absence of quantified impact. Most professionals underestimate how much of their work can be expressed numerically, and default to descriptive language that is both less compelling and less memorable than a specific number.
You are helping me quantify the impact of my
professional contributions for a performance review.
Here are the achievements I have described
in qualitative terms: [paste the relevant sections]
For each achievement, please:
1. Identify every dimension along which the
impact could plausibly be quantified:
time saved, revenue influenced, cost reduced,
error rate decreased, process accelerated,
team size managed, budget overseen,
stakeholders influenced, or any other
measurable dimension
2. Ask me the specific questions that would
give you the data to quantify each dimension
3. Where I cannot provide precise figures,
suggest conservative estimate ranges
I could use with appropriate qualifying language
Do not invent numbers. Do not suggest figures
that cannot be supported by reasonable estimation
from the information I provide.
The instruction not to invent numbers is not a formality. Language models will generate plausible-sounding statistics when asked to quantify impact without sufficient data. The explicit prohibition keeps the output grounded in what you can actually defend if someone asks where the number came from.
The manager's lens prompt
The most underused technique in performance review preparation is reading your own review through the eyes of the person evaluating it. Most professionals write reviews from their own perspective and forget that the reader has a different set of priorities, pressures, and evaluation criteria.
You are my direct manager reading my performance
review for the first time.
My manager's context: [their role, what they
are responsible for, what pressures they are
under, what they have told me they value,
any relevant organisational dynamics]
Here is my draft review: [paste draft]
Please give me your honest reaction as my manager:
1. What is the strongest part of this review
and why does it land well from your perspective
2. What is the weakest part and what does it
make you think or question
3. What is missing that you would want to see
before feeling confident about my trajectory
4. What single change would most improve
this review's effectiveness in making
the case I am trying to make
Be direct. A performance review that does not
work in the room does not serve either of us.
Running this prompt after completing the draft consistently surfaces at least one significant gap or framing problem that the writer cannot see because they are too close to the material. The manager's lens is the one that matters most and the one that writers are least able to inhabit without external help.
The development section
The development section of a performance review is where most professionals either damage themselves unnecessarily or miss an opportunity entirely. Listing genuine weaknesses without framing destroys the narrative. Listing non-weaknesses disguised as areas for growth fools no one and signals low self-awareness. The effective approach is precise and honest: identify real areas for development, demonstrate that you understand why they matter, and show evidence that you are already doing something about them.
You are helping me write the development section
of my performance review in a way that demonstrates
genuine self-awareness without undermining
the overall case I am making.
The development areas I am considering disclosing:
[describe them honestly, including how significant
they are and what you have done about them]
My role and what is expected at my level:
[brief description]
Please help me:
1. Assess which of these development areas
are worth including and which would be
better addressed in a direct conversation
rather than a written document
2. Frame each included area in a way that
leads with the action taken rather than
the gap identified
3. Connect each development area to a
specific capability that is relevant
to where I want to go next, rather than
just where I have fallen short
Constraints: Honest framing only.
No "my greatest weakness is that I work too hard."
No development areas that are actually strengths
in disguise. The reader will see through both
and it will cost more credibility than
the original gap would have.
A note on timing
These prompts are most effective when run three to four weeks before your review is due, not the night before. The twelve-question dialogue in particular benefits from time between the conversation and the drafting, because the act of articulating your year to a rigorous interlocutor tends to surface things you did not realise were significant until you said them out loud.
The professionals who use this framework as a genuine reflection exercise, rather than a drafting shortcut, consistently produce reviews that are more specific, more honest, and more strategically effective than those who treat it as a faster way to fill in a template.
The distinction between those two uses of the same tool is, in miniature, the distinction this newsletter has been making since Issue #1. AI amplifies the quality of the thinking directed at it. The thinking remains your responsibility.
That is sixteen issues. Eight weeks of the foundation this newsletter is built on. Starting Monday, we go deeper: into the specific industries, roles, and dynamics where the AI transition is moving fastest and where the gap between informed and uninformed professionals is widening most quickly.
The next eight issues will be the most practically useful we have published. The groundwork is laid. Now we build on it.
See you Monday.
— The Artificial Idea team
