Most of your developers, analysts and support staff are already using AI assistants, whether or not you have a policy. That changes who you should hire and how you should test them.
The skill people call prompt engineering is real, but it is narrower than the hype suggests. It is the habit of giving a model clear instructions, enough context and a way to check its work. Good prompting is a symptom of something else: the person understands the problem well enough to describe it.
What the tools now do for you
Assistants handle a lot of routine work well:
- Boilerplate and repetitive code
- Unit test drafts
- Documentation and summaries
- Data cleaning and formatting
- Searching internal knowledge bases for known errors
That reaches beyond developers. Data analysts use them to prepare data. Business analysts use them to pull together reporting. Support engineers use them to trace incidents faster. If you hire in any of those areas, your job descriptions should reflect how the work is done today.
What the tools still do badly
Models are weak where context is thin and stakes are high:
- Understanding your business, your customers and your regulations
- Debugging odd failures and handling edge cases
- Designing a system that will hold up in two years
- Deciding what should not be built
Those gaps are where your hiring budget should go. A developer who can accept a model's output without checking it is a liability. One who reads it critically, spots the flaw and fixes the design is worth more than before.
The junior hiring question
A lot of the routine work that used to train junior developers is now automated. Some leaders conclude they should only hire seniors. That is a short-term view. Senior people retire, move on or get poached, and you need a pipeline.
A better approach is to change what junior roles are for:
- Have them review and test AI-generated code, with a senior engineer reviewing their review
- Give them real exposure to production issues and customer problems early
- Pair them with someone who explains why a design was chosen, not only what it does
- Set clear rules on what can and cannot be pasted into external tools
Juniors who learn this way build judgment faster. Juniors left alone with an assistant can ship code they cannot explain.
How to screen candidates now
Stop testing memorized syntax. The assistant has that covered. Test what it cannot supply.
| What to test | How |
|---|---|
| Problem framing | Give a vague business request and ask what they would clarify first |
| Critical review | Hand over flawed generated code and ask them to find the problems |
| Architecture | Ask them to explain a trade-off they made and what they would change |
| Tool habits | Ask how they verify output and what they never delegate |
Let candidates use their own tools in an exercise if your team will use them on the job. Watch how they check the answer, not how fast they get one.
Where demand is likely to grow
When routine coding gets cheaper, the work around it grows. Expect steady need for people in enterprise architecture, cybersecurity, DevOps and data engineering. These roles depend on judgment, accountability and knowledge of how systems fit together. They are also hard to fill in both Canada and the US, so plan lead time.
If you hire for operations, finance or sales analytics, the same logic holds. Someone who can use AI to speed up reporting but still catch a wrong number is the person you want.
Set expectations in writing
Before you hire, decide a few things:
- Which tools are approved and what data may go into them
- Whether generated code must be reviewed before it is merged
- How you will measure output, so that volume of code is not mistaken for value
- Who owns the result when the assistant is wrong
Candidates with good judgment will ask about these. Their absence tells them something about your team.
What to do next
Rewrite one open job description this week. Drop the syntax checklist, add the judgment and review skills you need, and change your interview exercise to match. Then look at your junior pipeline and decide how those roles will teach people under the new conditions.
A Traverse recruiter can help you define these roles and find AI and data, technology and operations candidates who have the judgment to work well with the tools.