HR Cloud's AI agents now run on one source of company knowledge: the AI Knowledge Base. It's an AI knowledge base for HR that decides what every agent knows before it answers an employee, a manager, or a job applicant.
An HR agent is only as good as the knowledge behind it. Train it with your real policies and it gives real answers to your questions. No need to leave it guessing and giving it an excuse to make things up.
The AI Knowledge Base is where you curate what your agents can say, and set the rules for who can access which parts of it. Here's what it does.
You build the knowledge base out of Knowledge Spaces. Each space is a named collection, say a Company Handbook or a Manager Playbook, with three things attached:
You decide who has access to which space. Assign it to everyone, or limit it to a specific department, job role, or employment type, and combine those rules with AND and OR conditions when access needs to be precise.
That level of targeting is where most RAG-based HR systems stop. But here, the space lives inside HR Cloud, alongside People, your HR system of record, so you target audiences using the groups and org structure you already manage.
The space lives inside HR Cloud, alongside People, your HR system of record, so you can target audiences using the same groups and org structure you already manage.
Then you add the information that should enrich the space.
Upload your handbooks, benefits summaries, hiring guides, and procedures as doc, docx, pdf, or txt files. The agents answer from those documents; a question about parental leave returns the provisions of your policy not a generic web result.
Employees, managers, and applicants type the question the way they'd ask a colleague. The agent replies with an answer pulled straight from the space they're allowed to see.
Most HR knowledge is sensitive, so access has to be controlled. The AI Knowledge Base provides answers based on the permissions granted to the person raising the queries. Each person gets a response built only from the spaces their role and access allow, so a field technician and an HR admin can ask the same question and get answers according to their permission level.
Internal and applicant knowledge stay apart. A space built for job applicants can't reach into employee records or internal policy, so the Applicant Assistant can't accidentally reveal internal knowledge or employee information to job applicants.
Every time someone turns on an AI feature, HR Cloud records the consent with a name and a timestamp. The platform runs on SOC 2 Type II controls and supports GDPR, with an audit trail behind it.
Most vendors will tell you permission-based access is the hard part of an HR knowledge base. We built it in right from the start.
The same knowledge base feeds every agent you run.
The AI Assistant Bot answers employee questions from it.
Maya, the onboarding agent, uses it to guide new hires through their first weeks.
The Applicant Assistant draws on its applicant-only spaces to help candidates.
Update a policy once and every agent reflects it. We're rolling out deeper actions across the coming releases, so the agents will start completing tasks from the same knowledge, like filing a time-off request or updating an employee detail, while that knowledge stays in one governed place.
Think of the agent as the part that talks and the knowledge base as the part that knows. A capable agent sitting on thin or open-ended knowledge still gives confident wrong answers, and in HR a wrong answer about leave or eligibility carries expensive risk.
Grounding an agent in your own documents is what the industry calls retrieval-augmented generation, or RAG. The agent retrieves from your content before it answers. HR Cloud's version is permission-aware: what it pulls back depends on who's asking and what they're cleared to see.
Employee self-service portals have promised to fix this for years, yet people still searched, scrolled, gave up, and emailed HR anyway. SHRM's guidance on self-service shows how much routine work that leaves on HR's plate. A grounded AI knowledge base for HR closes that loop, because people get an accurate answer the moment they ask.
Grounding your agents in one knowledge base pays off in concrete ways. Fewer repetitive questions reach your team, answers come faster during hiring and onboarding, and HR can take on more work without adding more HR staff.
Deloitte's 2026 Global Human Capital Trends found nearly 60% of workers already use AI at work, yet only 14% of leaders feel adept at shaping how people and AI work together, which is exactly the gap a governed knowledge base closes.
How is an AI knowledge base different from an HR chatbot?
The chatbot is the part employees talk to. The AI knowledge base is what it reads before it answers. Without a governed knowledge base, a chatbot answers from generic training data. With one, it answers from your company's actual policies.
What should an enterprise AI HR knowledge base include?
Document grounding from your own files, role-based access so each person sees only what they're allowed, isolation between internal and applicant knowledge, recorded consent, and security controls like SOC 2 and GDPR.
How does AI keep HR answers accurate and compliant?
It answers only from the documents you upload and the spaces a person is allowed to see, and it keeps an audit trail of consent and access.
Can an AI knowledge base respect permissions and sensitive data?
Yes. Every answer is built from the spaces a person's role allows, and applicant-facing knowledge is kept separate so it can't expose internal employee data.