Nobody wants to learn seven HR systems. They want to ask one question.
HR AI Assistants is a suite of seven specialist assistants sitting behind a single agentic orchestrator. Employees and managers ask in plain language; the orchestrator reads intent, routes to the right specialist invisibly, and returns a grounded, action-ready answer — talent matching, recruitment tracking, policies, leave, expenses, performance and product know-how, from one conversation.
HR AI Assistants — Agentic HR Assistant Suite
The problem underneath.
The failure mode of HR self-service is well known: the portal has everything, and nobody can find anything. So the questions come back to the HR team by email, one at a time, and the portal that was supposed to save time becomes an additional system to maintain.
The fix is not a better menu. It is removing the menu entirely. An employee asking "how many days do I have left and can I take the 14th" should not have to know that leave balance and leave request live in different modules — and they should certainly not have to pick which assistant to talk to. The orchestrator handles that invisibly, asks one clarifying question when genuinely needed, and handles multi-domain requests in sequence.
Two constraints make this safe to deploy. Every write action — an approval, a rejection, a submission — requires explicit user confirmation before anything is committed. And every documentation answer is grounded through RAG on your own library and cites the exact manual, section and page. The assistant does not improvise about your policy, and it does not act on your behalf without being told to.
Input, decision, output.
The full path a request takes through HR AI Assistants — nothing hidden in the middle.
Plain-language question
An employee or manager asks in natural language — no module, no menu, no form.
Agentic orchestrator
Reads intent, routes to the right specialist, asks one clarifying question when needed.
Grounded answer
Policy and documentation answers cited to the exact source manual, section and page.
Confirmed action
Leave, expense and approval actions executed only after explicit user confirmation.
Manager view
Performance, objectives, EPR interviews and development actions coming due or overdue.
What it actually does.
Talent Assistant
Reads an uploaded CV, matches it against every active job posting and ranks the top five roles with a transparent 0–100 score.
Recruitment Assistant
Keeps hiring managers on top of open tasks, direct reports and incoming new joiners as clickable, ready-to-act links.
Policy Assistant
Surfaces the right HR announcement or policy on demand, rendering full rich-text content and never truncating the answer.
Leave Assistant
Files leave requests, checks conflicts and lets managers approve or reject — single or in bulk — always with explicit confirmation.
Expense Assistant
Handles expense records and approvals end to end, with clear status states and clickable links back into each record.
Performance Assistant
A manager-grade view of evaluations, objectives with KPIs, EPR interviews and development actions coming due.
User Guide Assistant
Searches hundreds of HumanSoft PDF manuals via Azure AI Search and answers in plain language, citing guide, section and page.
Agentic Orchestrator
The central routing layer — users never pick an assistant, and multi-domain requests are handled in sequence.
Who this is for — and who it isn't.
Most vendors only answer the first half. The second half saves everybody a quarter.
A good fit if
- Your HR team answers the same questions repeatedly by email despite having a portal
- You run HumanSoft or a comparable HR suite with real data behind an API
- Managers avoid the system for approvals because the workflow is tedious
- You have a large PDF documentation library nobody reads
Probably not if
- Your HR data lives in spreadsheets rather than a system with an API
- You need the assistant to make hiring or performance decisions — it surfaces, it does not decide
- Your headcount is small enough that HR knows every question by name
Questions we get asked.
Do users choose which assistant to talk to?
No, and that is the point. The orchestrator reads intent and routes invisibly. A user asking about a colleague's leave and their own expense claim in the same message gets both handled in sequence, without ever knowing two specialists were involved.
Can it approve leave without me?
No. Approvals, rejections and submissions always require an explicit user confirmation before anything is written. The assistant prepares the action and states exactly what it will do; you confirm.
How accurate are the documentation answers?
Documentation answers are grounded through RAG on your own manual library and cite the exact source — guide, section and page. If the answer is not in the library, the assistant says so rather than generating one.
How does CV matching produce a score?
The Talent Assistant reads the uploaded CV, compares it against every active posting, and ranks the top five with a transparent 0–100 score that can be traced back to the matching criteria. It is a shortlisting aid for a human recruiter, not a screening decision.
Does it work with HR systems other than HumanSoft?
The suite is built on HumanSoft and is deepest there. Integration with other HR platforms is possible through the function-calling layer, scoped case by case.
Often deployed alongside.
Lead Transform
Lead Transform runs a leader's growth journey end to end.
Ethos
Ethos semantically analyses internal email correspondence across a chosen date range and surfaces the messages that matter — rising tension, cultural misalignment, anger, and content that carries genuine ethical risk.
Sales Mentor
Sales Mentor is a coach agent that talks — by voice — with sales and marketing teams before and after meetings and activities.
See HR AI Assistants on your own data.
Start with a free 90-minute assessment. We map where this fits in your organization, what it would touch, and what it would return — then you decide.