Assessment
What the assessment includes
What the quick questionnaire checks, what the deeper interview asks, and why the assessment measures behavior instead of tool knowledge.
The assessment has two parts. The quick questionnaire gives the first orientation. The deeper interview verifies what someone actually does with AI.
The two parts complement each other. The questionnaire shows how someone perceives their own work with AI. The interview adds concrete examples, outputs, and working habits.
#Quick questionnaire
Spark takes about 2 minutes. The base public flow has seven questions:
| Question | What it looks for |
|---|---|
| Frequency | How often you use AI at work, including an explicit AI-first option |
| Tools | Which AI tools you actually use at least occasionally today |
| Use at work | How you use AI: for concrete tasks, with your own materials or data, or when making decisions |
| Things you have built | What you have prepared or built with AI: a reusable prompt or process, a contextual assistant, a tool connection, or your own service |
| Goals | What you most want to improve with AI; you select every option that applies |
| Blockers | What is holding your work with AI back today; you select all options that fit |
| Work context | Whether you are an employee, lead a company, serve clients, create your own products, study, or have another situation |
If you select employee in the public flow, an eighth question asks for a broad department category. Students and other non-employee options finish on question seven. In a company program, work context is skipped and department is the seventh question. You select it yourself; the company does not supply or prefill it for you.
Both middle questions allow multiple selections, and each requires at least one answer. If nothing applies yet, you can choose the separate “Not yet” or “Nothing yet” option.
The questions are not a quiz. A more advanced brand or a longer tool list does not automatically mean a higher result. The first Spark score uses only frequency and your answers about practical use and things you have built. Goals, blockers, work context, and department help tailor the report and recommendations, but do not add points by themselves.
Spark deliberately does not ask separate questions about output quality, iteration, or verification. Without concrete examples, those answers would only be self-assessment. Deep is where those signals are examined.
#Deeper interview
The deeper interview is led by Aimee and takes about 15 to 20 minutes. It asks like a curious colleague, not an examiner.
It typically asks about:
- the last concrete situation where you used AI,
- repeated tasks you do again and again,
- tools you actually use,
- outputs, templates, or workflows you created with AI,
- places where work with AI gets stuck.
At the end, there may be a practical scenario from your field. We are not looking for the "right answer". We want to understand how you would think about the situation and where you would involve AI.
#Why the assessment is short
With AI, a few good questions are more useful than a long questionnaire. One concrete situation often says more than ten general scales.
That is why we combine a short questionnaire with an interview. When what someone says about themselves differs from what they describe in concrete examples, concrete behavior carries more weight.
#What the assessment does not test
- Knowledge of specific prompts.
- Awareness of the latest AI trends.
- Technical programming skills.
- Personality or "AI talent".
We measure current practice: how someone uses AI, what outputs are created, and whether the way of working changes.
#How answers become a result
The questionnaire creates the first low-confidence hypothesis. The interview confirms, adjusts, or sharpens it. The result is a level, a natural style of working with AI, a next step, and a baseline numeric measure for later comparison.
If a new questionnaire or scoring version is released between attempts, the old report stays intact. A larger methodology change may create a new baseline. The current split of one question into two preserves the weights of shared answers, so the same selections produce a comparable Spark result.