Verification
How do you know it worked?
Checks of the actual result, tests, review, and evidence that supports the claim of success.
Ask them
Show a time the AI said it was done and your checks found otherwise.
Runpoint AI work assessment
Show your work.
Radically better businesses need people who can put AI to work. Look at something they built, the decisions they made, and how they checked the result. That gives you a useful place to start.
The assessment opens on Overflow. Choose your sources, run it in your own coding agent, and review the report before deciding whether to share it. Your AI tool's data terms apply.
01What we look for
The habits behind useful work.
At Runpoint, we want people who can carry a business problem into a working solution and stay responsible for the result. This guide adapts Overflow's evidence approach into five things you can discuss with a candidate or teammate.
Judgment
What did you decide, and why?
Choices about scope, trade-offs, risk, and when to bring in someone with different expertise.
Ask them
What did you change or decline to build, and what led you to that decision?
Communication
Could someone else carry the work forward?
Clear instructions, useful corrections, decisions others can understand, and a handover tied to the intended result.
Ask them
Show the original request, a correction you made, and what the next person received.
Working with agents
How did you direct and review the AI?
Breaking work into useful parts, supplying context, reviewing outputs, and recovering when an agent goes off course.
Ask them
Which parts did you do yourself, which did the AI do, and how did you check the difference?
Follow-through
What happened after the first version?
Responsibility for getting the work used, fixing problems, and supporting it after the initial output.
Ask them
Was this an experiment, a working prototype, or something people used? Show what establishes that.
Keep the evidence and its limits together. A working prototype, something used by customers, and a system someone maintains are different kinds of experience.
02Use it yourself
Bring better evidence to the conversation.
Hiring
Ask for a relevant piece of work.
Agree on the role, ask the candidate to choose examples they can share, and use the same review questions for people applying to that role. Discuss the decisions and verify the claims that matter.
Team development
Find the next thing to practice.
Choose examples with the employee. Look for a habit to strengthen, agree on the next piece of work, and revisit it together. Missing evidence is a question to explore.
Self-review
See what your work demonstrates.
Run the assessment or review selected examples yourself. Separate what you can show from what you want to learn next, and build a useful record of your progress.
Keep the person involved.
The person chooses the material and reviews what they share. If their work does not produce coding-agent logs, use relevant artifacts or a work sample. Access to tools, activity volume, and a partial history cannot tell you everything about someone's ability.
Use the findings alongside a conversation and the requirements of the job. This guide provides no overall score, automatic hiring decision, or employee rating.
03Get the tools
Make it useful in your company.
The Runpoint review kit includes an evidence worksheet, five discussion prompts, a development plan, and an optional prompt for organizing selected evidence with AI. No signup required.
Download the review kit ↓A Markdown file you can open in a text editor or paste into your approved AI tool. The download does not collect assessment results.
From Overflow
Start with the full AI Work Assessment.
Overflow turns selected work history into a report you can review. Our kit helps you discuss the evidence in the context of a role or a development goal.
Open the original assessment ↗