What Happens to HR When Employees Work With AI Agents?
By [Author name] · [Month 2026]
What Happens to HR When Employees Work With AI Agents?
Read as
You're reading the key passages · 2 min
These are the author's own words, unchanged. Mimansa editors picked the passages, and · · · marks where parts are left out. The full article adds the detail behind each change: junior hiring, benchmarking, AI budgets, the employee journey and how agents should behave.
You're reading the full article · 4 min
Parts that are not in the key passages have a line beside them, like this one.
Kyun · your host
Today's question: what happens to HR when employees start working with AI agents? Over to the author.
Fig. 1 · A new name on the desk
Humans tend to overestimate the short-term impact of a new technology and underestimate its impact over time.
That is worth remembering when we talk about AI and jobs. Predictions range from the disappearance of entry-level jobs to a future where people no longer need to work.
When predictions differ so widely, how should organisations prepare?
I will make three assumptions:
AI capability continues to improve, but we do not reach AGI in the short term.
AI becomes more affordable, so more people and companies use it.
Governments do not restrict its use so heavily that widespread workplace adoption stops.
If these assumptions hold, I think HR will need to focus on four changes.
Kyun
Four changes, one at a time. Which one is closest to your desk today?
Change 1 of 4
1. Organisation design will change
· · ·
As this happens, organisations will need fewer people doing operational coordination. I expect more roles to move closer to customers, where employees spend time understanding needs, building relationships and taking responsibility for resolving problems.
But this movement will not happen automatically. Someone who has spent ten years managing an internal process will need support to succeed in a customer-facing role. HR will have to work out which roles disappear, which change and how people prepare for the work that remains.
· · ·
There is also a longer-term problem. If agents take over the work juniors used to do, businesses may stop hiring them.
· · ·
Otherwise, where will experienced people come from in five or ten years?
Change 2 of 4
2. Roles will need fresh evaluation and benchmarking
If AI takes over a significant part of a role, its current job description, grade and salary benchmark need to be questioned.
· · ·
HR will need to evaluate what each role has become, including the judgment, responsibility and decisions that remain with the employee.
· · ·
Performance assessment will need to account for those resources. If two employees have very different AI budgets and tools, how fairly can we compare their output?
Change 3 of 4
3. HR will need to define how humans and agents work together
Employees will increasingly delegate work to agents. They will need to explain the task, set boundaries, check the output and decide when to intervene.
· · ·
How do you assess someone’s potential when they have an external brain?
· · ·
Performance expectations must make accountability clear. An employee who delegates work to an agent still needs to check that it meets the required standard.
Kyun
When an agent joins a team, who briefs the people and who owns the agent? The playbook starts there.
When I receive an email from a colleague, will I know whether they wrote it, reviewed it or even saw it?
· · ·
AI is changing how much we trust the work we receive from other people.
Agents communicating on our behalf will make this harder. One employee may configure an agent to chase people aggressively. Another may allow it to make commitments without checking. Both agents represent the company.
Kyun
Now the author turns from the four changes to HR itself.
· · ·
HR cannot address these questions without understanding how AI works and using it in its own work.
· · ·
The question it leaves
How will HR judge which roles should disappear, which deserve greater value and which must be protected for the future? And if HR cannot make those judgments, who will make them on employees’ behalf?
Humans tend to overestimate the short-term impact of a new technology and underestimate its impact over time.
That is worth remembering when we talk about AI and jobs. Predictions range from the disappearance of entry-level jobs to a future where people no longer need to work.
When predictions differ so widely, how should organisations prepare?
I will make three assumptions:
AI capability continues to improve, but we do not reach AGI in the short term.
AI becomes more affordable, so more people and companies use it.
Governments do not restrict its use so heavily that widespread workplace adoption stops.
If these assumptions hold, I think HR will need to focus on four changes.
Kyun
Four changes, one at a time. Which one is closest to your desk today?
Change 1 of 4
1. Organisation design will change
A lot of operational work involves moving information between people and systems: checking information, following up, scheduling meetings and updating records.
AI agents can take on more of these tasks because they can use tools and work across software.
As this happens, organisations will need fewer people doing operational coordination. I expect more roles to move closer to customers, where employees spend time understanding needs, building relationships and taking responsibility for resolving problems.
But this movement will not happen automatically. Someone who has spent ten years managing an internal process will need support to succeed in a customer-facing role. HR will have to work out which roles disappear, which change and how people prepare for the work that remains.
HR’s own structure will change too. As agents handle routine queries, interview scheduling, follow-ups and system updates, HR teams will become smaller. HR business partners will need to spend more time on business decisions, organisation design and manager capability.
There is also a longer-term problem. If agents take over the work juniors used to do, businesses may stop hiring them.
HR and the CEO will have to protect the long-term talent pipeline. That means deliberately creating opportunities for juniors to shadow experienced colleagues, practise making decisions and learn from mistakes.
Otherwise, where will experienced people come from in five or ten years?
Kyun
Hold on to that question about juniors. It comes back at the end.
Change 2 of 4
2. Roles will need fresh evaluation and benchmarking
If AI takes over a significant part of a role, its current job description, grade and salary benchmark need to be questioned.
A role may have commanded a premium because it required specialist knowledge, extensive analysis or years of experience. What happens when an agent can perform much of that work?
Other roles may become more demanding because one employee can now manage work that previously required an entire team.
HR will need to evaluate what each role has become, including the judgment, responsibility and decisions that remain with the employee.
Benchmarking will become harder. Two companies may use the same job title while expecting very different contributions because their use of AI is different. Comparing titles and years of experience will tell us even less than it does today.
An AI budget may also become part of a role’s cost, alongside salary. Employees will need access to models, tools and tokens to deliver what is expected of them.
Performance assessment will need to account for those resources. If two employees have very different AI budgets and tools, how fairly can we compare their output?
Change 3 of 4
3. HR will need to define how humans and agents work together
Employees will increasingly delegate work to agents. They will need to explain the task, set boundaries, check the output and decide when to intervene.
We already train managers to delegate and give feedback to people. Employees will need to learn how to manage agents too. The skills overlap, but the behaviour is different. An agent can misunderstand an instruction, confidently produce an incorrect answer or fail to retain feedback across conversations.
This changes the employee journey.
During hiring, should candidates demonstrate what they can do with agents? Should we also assess what they understand without that support?
How do you assess someone’s potential when they have an external brain?
Induction will need to explain which agents already work in the organisation, what they can do and who is responsible for them.
Performance expectations must make accountability clear. An employee who delegates work to an agent still needs to check that it meets the required standard.
And when employees leave, someone must take ownership of the agents they built and understand the work those agents continue to perform.
Kyun
When an agent joins a team, who briefs the people and who owns the agent? The playbook starts there.
When I receive an email from a colleague, will I know whether they wrote it, reviewed it or even saw it?
We already see people sending AI-generated work they have barely checked. It looks finished, but the recipient has to verify the claims, find the mistakes and work out what the sender actually thinks.
AI is changing how much we trust the work we receive from other people.
Agents communicating on our behalf will make this harder. One employee may configure an agent to chase people aggressively. Another may allow it to make commitments without checking. Both agents represent the company.
Both agents represent the company.
HR will need to translate organisational values into clear expectations for agent behaviour. What can an agent commit to? When must it disclose uncertainty? When must it ask a human?
Trust also affects engagement. If an agent does the work an employee used to take pride in, what contribution will their manager recognise?
Kyun
Now the author turns from the four changes to HR itself.
HR cannot address these questions without understanding how AI works and using it in its own work.
Decisions about headcount, job value, accountability and careers are already part of HR’s responsibility. As AI changes those decisions, HR needs enough knowledge to challenge proposals and explain the consequences.
Fig. 3 · Who weighs the role?
The question it leaves
How will HR judge which roles should disappear, which deserve greater value and which must be protected for the future? And if HR cannot make those judgments, who will make them on employees’ behalf?
Kyun
Now you. Where do you land?
Tap one to see how other readers answered.
Kyun
Those were the key passages. The full article walks through each of the four changes in more detail.
Use this at work · Playbook
Bring Agents Into a Team
Bring an agent into a team with clear tasks, permissions, owners and a 30-day check.
