It is Monday, 9:12 a.m. HR has 31 unread Slack messages.
Fourteen of them are some version of the same question: is payroll still the 25th, or did it move for the holiday?
The answer is in the payroll calendar. The payroll calendar is in a PDF. The PDF is in a folder nobody has opened since March.
That is the job an HR chatbot is good at, and it is a smaller, more useful job than most of the AI for HR noise suggests. It answers the question. It does not make the decision. We will get to why that line matters.
The Questions That Eat an HR Team's Week
Ask anyone in people operations what fills their inbox and you will not hear "complex employee relations cases." You will hear leave balances.
Leave, payroll, and "where is that policy"
How many leave days do I have left. Is the office closed on the 4th. What is the limit for client dinners. Can I carry over unused days.
None of these are hard. Every one of them has an answer somebody already wrote down. They still arrive one at a time, in Slack, in email, and occasionally in the corridor, and each one pulls a person away from work that actually needs a person.
HR automation usually starts with forms and approvals. The questions around those forms get left behind.
The handbook is a PDF nobody searches
Most companies do have a handbook. It is 60 pages long, it was last updated in a hurry, and it lives on an employee self service portal that people log into once a year to download a payslip.
So employees ask a person instead. Not because they are lazy. Because asking is faster than searching.
An HR chatbot for policy and leave questions turns that PDF into something people can actually ask. The answer comes back in one sentence, with the page it came from attached.
Onboarding repeats the same ten answers
Every new starter asks where the laptop comes from, when the first payday is, and who approves their leave. Every manager answers them again.
Ten answers, repeated for every hire, is not onboarding. It is a very expensive FAQ.
The Candidate Experience Starts With a Reply
Most recruitment automation is built for the recruiter's side of the process. Candidates ask different questions, but the problem is identical: someone is waiting on a person who already knows the answer.
"Did you get my application?"
The most common candidate complaint is also the simplest to fix. Onrec's own feature on improving the candidate experience quoted it directly: applications that are never acknowledged, so candidates have no idea whether they are moving forward or not.
Then comes the next wave. When is the interview. Can I move it. When does the offer letter land. Each one is a small moment where a good candidate experience is won or quietly lost.
Where a recruitment chatbot helps, and where it should stop
It is good at the logistics. Confirming an application arrived. Explaining the next stage. Answering what the working pattern is, if the job ad already says it.
Interview scheduling automation is the natural next step, and a recruiting chatbot can offer candidates the available slots without three emails back and forth.
Where it should stop is judgment. It should not tell a candidate they look like a strong fit. It should not guess at salary if the range is not published. And it should never be the thing deciding who moves forward, which brings us to the part of AI in recruitment that has a legal edge.
Answering Is Not Deciding
There is a lot of enthusiasm for AI in HR right now, and plenty of vendors describe AI in human resources as if answering and deciding were the same thing. They are not, and the law has started to notice.
What NYC Local Law 144 covers
New York City's Local Law 144 regulates automated employment decision tools: software that substantially helps decide who gets hired or promoted, often by scoring or ranking candidates.
Under the law, an employer cannot use one unless it has had a bias audit within the past year, a summary of the results is publicly available, and candidates or employees have received the required notices. Enforcement began in July 2023, and it was the first US law to require independent bias audits of AI hiring tools.
Why a chatbot that only answers stays on the safe side
An HR chatbot that tells a candidate when the offer letter will arrive is not making an employment decision. Neither is one that explains the leave policy.
That is the line worth drawing before you buy anything. Let the tool answer questions. Keep the decisions with people, and keep any tool that does influence decisions under the audit rules that apply to it.
It also makes the internal conversation easier. Nobody on the team worries about being replaced by something that looks up the payroll date.
What an HR Chatbot Actually Does Well
Take away the hype and the job is narrow. That is a good thing.
Policy answers with the source attached
Every answer should show the page it came from. When the answer is wrong, the citation tells you which policy page is out of date. When it is right, the employee can check it themselves. Either way, nobody is arguing about what the chatbot "thinks."
Handoffs that arrive with the whole conversation
Some questions need a person. A disputed payslip. A sensitive leave request. A complaint about a manager.
When those arrive, HR should get the full conversation, not a message that says "an employee has a question." An AI HR assistant that hands over cleanly is worth more than one that tries to answer everything.
Coverage when managers ask from another time zone
The manager in San Francisco asks about the notice period at 6 p.m. It is 2 a.m. in London, where HR sits. Nobody is awake.
A chatbot answering from the same policy pages, in the language the manager writes in, means the question gets answered tonight instead of tomorrow afternoon. An employee service center that closes at 5 p.m. stops being the bottleneck for the easy questions.
Choosing an HR Chatbot Your Team Will Trust
Most lists of AI tools for HR compare features, and most coverage of AI in recruitment does the same. We would compare behavior.
Ask it the questions that should fail
Load your handbook, then ask about a policy you do not have. The right answer is some version of "that is not in our documents, let me get someone." A confident answer to a question your handbook does not cover is the biggest warning sign there is.
Check where it lives
If your team lives in Slack, the chatbot should too. An employee self service tool that needs its own login will get exactly the attention your current portal gets.
Check what it does not do
Good HR automation software is clear about its limits. If a vendor pitches one of its AI in recruitment platforms as able to score or rank candidates, you are looking at a different category of tool, with the audit obligations above attached.
How Agentency Handles an HR Chatbot
We build one of these, so here is how our own platform lines up against that list.
It answers from your handbook, with the source shown
Agentency answers people operations questions from the policy material you give it: crawled pages, uploaded PDFs, or pasted text. Every answer shows the source it came from, and the built-in test chat lets you ask the questions your team is tired of before any employee sees a reply.
It checks live records through Call Actions
Some answers are not in a document. A leave balance, for example. Through a Call Action, the agent can look up a live record from the API of the HR system you already run, instead of guessing from last quarter's spreadsheet.
Higher risk actions can be set to require an explicit confirmation before anything runs.
It declines and hands HR a complete brief
When the retrieval quality gate finds nothing relevant, the reply is blocked. The agent says it does not know, captures the employee's question, and hands it to HR with the full transcript attached. That is the AI HR assistant behavior worth paying for.
It lives in Slack and the channels your team already uses
The same agent answers in Slack, on a web widget, in Telegram, or through a custom API, all from one knowledge base. It replies in the language the person writes in, which covers the manager in San Francisco and the new starter in Pune without a second setup.
It shows you which answers are missing
Conversations show which questions the policy pages could not answer, so you fix the handbook rather than arguing with the model.
Behind all of it, data is encrypted in transit, each workspace is isolated, and two-factor authentication is available on every plan.
Frequently Asked Questions
What is HR automation?
It is the use of software to handle repetitive HR work, such as leave requests, onboarding checklists, payroll reminders, and policy questions, so the team can spend its time on work that needs judgment. An HR chatbot covers the questions that surround those processes.
What is candidate experience?
It is how an applicant experiences your hiring process, from the job ad to the offer or the rejection. Speed and clarity of communication shape most of it, which is why unanswered questions do so much damage.
How to improve candidate experience
Acknowledge every application, tell candidates what happens next and when, and answer logistics questions quickly. A chatbot can handle the logistics around the clock, which leaves recruiters free for the conversations that actually need them.
Will HR be replaced by AI?
Not by an HR chatbot. It takes the payroll-date questions off the team's plate. Employee relations, difficult conversations, and hiring decisions stay with people, and in New York City, tools that substantially influence hiring decisions come with audit requirements of their own. The realistic outcome of AI for HR is fewer interruptions, not fewer HR people.
How to use AI in recruitment
Start with an AI chatbot for recruitment that answers what candidates ask most: application status, interview logistics, and next steps. Keep screening and hiring decisions with your recruiters, or make sure any tool that influences them meets the audit rules where you hire.
What This Comes Down To
The payroll date was never the hard part of HR. It was just the part that arrived fourteen times on a Monday.
An HR chatbot takes those questions off the team, answers them from the pages you already wrote, and passes the rest to a person with the whole conversation attached.
Let it answer. Keep the deciding for people. That is the whole playbook, and it fits on one page, which is more than most handbooks can say. HR automation, done this way, gives the team its Mondays back, and that is the most honest promise AI for HR can make.
Agentency is an AI chatbot builder for businesses that want every customer question answered, day and night, without the guesswork. Teams add the documents and pages they already maintain, and the Agentency agent answers from that material only, with the source shown. When the answer is not there, it says so and hands the conversation to a person with the full transcript attached. One agent covers the website widget, WhatsApp, Instagram, Messenger, Telegram, Slack, and more than ten channels in total, and it replies in the visitor's language. Businesses in ecommerce, real estate, restaurants, IT support, and internal knowledge use Agentency to answer, resolve, and sell around the clock.
