AI agent for HR teams
An agent that handles the routine side of HR: it answers staff questions from your policies, collects onboarding documents, runs leave and expense requests up to the manager's approval, tracks contract dates and sorts applications for a person to decide.
- Answered 9 staff questions about leave rulesDone
- 3 new hires reminded about missing documentsDone
- 2 contracts end next monthAsks a person
- Short list for the sales role ready for reviewAwaiting approval
An example. Your agent is built around your own work.
How it works
-
Answer from your policies
Employees ask in Teams, Slack, email or WhatsApp about leave, payday, expenses or equipment. The agent answers from your approved HR documents, quotes the policy it used, and checks the person's own balances in the HR system.
-
Run the routine request
For a leave request or an expense claim, the agent collects what is missing, checks the balance and the rules in your HR system, and sends the request to the manager for approval.
-
Chase documents and dates
It sends a new hire the onboarding forms, reminds people of what is still missing, and watches contract ends, probation reviews, permit expiries and training deadlines for HR.
-
Pass on what needs a person
Pay disputes, medical leave, complaints and anything about discipline go straight to an HR person with a summary. For applications, the agent prepares a short list against your criteria; people decide.
New to AI agents? How they work, in plain words
Real examples, setup steps and tool connections for this agent
An example conversation
An example of how a conversation runs. The grey lines show what the agent does in your systems meanwhile.
- CustomerHi, how many days of annual leave do I have left? I want to take the last week of August.
- Identified the employee through company sign-in: Leila Karimi, logistics team. Read her leave balance in the HR system: 9 days left. Checked team calendar: 2 other people already off 26 to 30 August.
- AgentYou have 9 days left. The last week of August is 5 working days, so you have enough. Two colleagues in logistics are already off that week, which your manager will need to approve. Shall I send the request?
- CustomerYes please.
- Created the leave request for 26 to 30 August in the HR system. Sent an approval card to her manager on Teams with the team calendar attached.
- AgentSent to your manager. I will tell you here as soon as they answer.
- CustomerOne more thing: I think my last payslip had the wrong overtime. Can you fix it?
- AgentI cannot see or change pay corrections, and a possible error in your payslip is something HR needs to look at. I have passed it to the payroll team with your name and the month, so you do not have to explain it again.
- Opened a case for payroll with the employee's message and the payslip month. Assigned it to the payroll queue. Did not read the payslip figures.
How it is built
- Your policies are the only source
- We load your HR documents (leave rules, expense policy, onboarding checklist) and keep each version marked with the date it took effect. Every answer shows which document it came from, and an old version stops being used on the day it is retired.
- Each person sees their own record
- Employees sign in through your company single sign-on, and the agent can read only that person's data. Managers see only their team. These limits are enforced in the HR system's API, not left to the model's good behaviour, and we test them with accounts from different teams.
- A language model understands the question, rules decide the outcome
- A language model reads the question and finds the right policy text. Rules decide what happens next: whether the balance covers a leave request, who the approver is, which topics always go to HR.
- Write actions need approval
- The agent can prepare a leave request, an expense claim or an equipment order. Before anything changes in the HR or payroll system, the manager approves it, and the agent checks that the write succeeded before telling the employee.
- Sensitive topics go to people
- Pay disputes, medical leave, discipline and complaints are set to hand over to HR at once, with a short summary. The agent does not answer them itself, and it never changes pay or status.
- Applications are sorted, not decided
- For hiring, the agent compares applications with the criteria you set and explains the match. The list is a sorting aid for the recruiter. Selection and rejection are made by people.
- Logs and privacy
- Questions, unanswered questions, approvals and corrections are logged for audit, and the list of unanswered questions shows which policy text is missing. Chat logs have a retention period you set, and the agent can run on your own servers.
What you can connect to it
- HR systems (BambooHR, Workday and others)
- Reads employee records, leave balances and onboarding tasks, and creates approved time-off requests.
- Payroll software (Gusto, QuickBooks Payroll and others)
- Reads pay dates and schedules to answer questions, and passes approved changes to the payroll team without altering pay itself.
- Microsoft Teams
- The place where employees ask, and where managers get and answer approval cards.
- Slack
- The same for companies that work in Slack.
- Outlook and Gmail
- Sends welcome emails and forms to new hires, and collects returned documents.
- Google Calendar and Outlook Calendar
- Books onboarding sessions and training, and checks who is off before a leave is approved.
- Shared drives (SharePoint, Google Drive)
- Holds the policy documents the agent quotes and the documents employees return.
- Applicant tracking and job boards
- Reads incoming applications and CVs, compares them with your criteria and attaches the reasons.
- Google Sheets or Excel
- Holds contract-end dates and checklists if you do not yet have an HR system.
Connects to your tools. You stay in control.
- Works with what you have
- Your HR or payroll system (BambooHR, Workday, Gusto and others), Microsoft Teams, Slack, Outlook, Gmail, calendars, shared drives, job boards, spreadsheets.
- A person approves what matters
- Payments, orders, discounts or anything you choose wait for approval. Every action is logged.
- Can run on your own servers
- The agent, its data, its logs and even the language model (open models such as Llama, Qwen or Mistral) can run on your servers, so documents and customer data stay inside your network. How that works
What it does not do
- It does not make hiring or firing decisions. It sorts and explains; a person decides.
- It does not approve leave, change pay or change an employee's status. Managers and HR do.
- It does not answer pay disputes, medical leave or misconduct questions. Those go to an HR person.
- It does not show an employee anyone else's data, and it answers from the approved policy text, not from guesses.
- It does not keep personal data in its own memory. Records stay in your HR system.
Questions
Does the agent make hiring decisions?
No. It sorts applications against your criteria and explains why, and a person reviews and decides. We recommend checking the criteria themselves for bias, since the agent applies exactly what you set.
What about personal data?
The agent can run on your own servers, sees only what you allow, and every access is logged. Each employee can reach only their own record, and the rule is enforced in your HR system's access controls.
What can an AI agent do for HR in practice?
It answers repeated questions from your policies, runs routine requests such as leave, expense claims and equipment orders through your HR system, sends and collects onboarding documents and tracks key dates. It does not take decisions about people.
What if the policy text is out of date and the agent gives a wrong answer?
Each answer is tied to a document version and shows its source. When a policy changes, we replace the text and retire the old one, and the unanswered-questions list tells HR where the documents are missing.
How do we build an HR agent without disrupting HR?
Start with questions only, then add write actions after HR has seen how it answers. Pick two or three frequent, rule-based requests, connect your HR system, add single sign-on and test with real accounts from different teams before launch.
Can it automate HR processes such as onboarding?
Yes. It sends the forms and checklist, books the sessions, reminds people of missing items and updates the record. Steps that need a decision, such as pay or contract terms, stay with HR.
Do we need an HR system for this?
No. If you use spreadsheets, the agent can work with them for contract dates and leave records, and it can be moved onto an HR system later.
How much does an agent like this cost?
It depends on the task and the systems it connects to. After a first call we give you a clear estimate of cost and time.
Want this agent built around your work?
Tell us in a few lines what the job looks like today and we will set up an online intro call.
Request sent
We will contact you to set up an intro call.