Tier-1 HR Query Assistant

Where it works well

It deflects the same dozen questions asked every Monday, with the clause shown so the asker trusts the answer.

Best for a People & Culture team of one to a handful in an org of a few hundred staff, where the same person fields the routine and the sensitive.
It earns its keep only when the volume is real and the answers genuinely live in approved documents — an enterprise agreement, a leave policy, payroll SOPs, a handbook.
The hours it deflects go back into the grievances, edge cases and judgement calls that actually need a person, not into cutting the role.

Most of an HR inbox is the same handful of questions in slightly different words — personal/carer's leave accrual, when payday falls, notice periods, what salary-sacrifice arrangements exist. None are hard; they're just answered from documents your team already maintains, and answering them one by one is invisible work that never shows up as a project.

Where it works badly

It is confidently wrong when the corpus is stale — and the cited answer looks more authoritative than the document behind it.

It works badly anywhere the answer isn't written down — if your "policy" lives in a manager's head, an old email, or custom and practice, it can't retrieve it.
At best it refuses; at worst it answers from an adjacent document that doesn't quite apply, which reads as authoritative but isn't.

The honest test

Pick your ten most common HR questions and ask whether each has a single, current, written, approved answer you could point to a clause for. If most resolve to "it depends, ask me", the real first job is writing the policy down — not buying an assistant.

Feed it last year's enterprise agreement after a new one is ratified, or a policy that has since been reissued, and it will quote a superseded clause with full confidence, because it has no way to know the document is stale. That is the trap.

What it doesn't do — and shouldn't

It surfaces what your documents say and cites where. A person decides what the answer should be. That boundary is deliberate.

WHAT IT DOES

Quotes the exact clause it relied on — for example a leave policy section or an enterprise-agreement clause

Shows a grounding-confidence score and the passages the answer was built from

Flags a conflict when two documents address the same question differently, returning both cited

WHAT IT WON’T

Answer a staff member's specific circumstances — a disputed carer's-leave claim, a redundancy or a grievance

Decide which document governs when the agreement, a policy and the handbook disagree

Invent an answer when the topic is absent from the approved corpus

HR and payroll answers carry legal and financial consequence under the Fair Work Act 2009, the National Employment Standards and the modern award that covers your staff, with the Fair Work Ombudsman as the regulator. The worst outcome isn't "no answer" — it's a confident wrong one a staff member acts on. So it refuses loudly and routes to a human on anything outside the corpus, and keeps a person on every flagged answer.

What your data has to look like

Approved, versioned documents specific enough to answer from, with a named owner and a re-index step.

Typical readiness

across orgs we see, before the first job

Enterprise agreement / award coverage, with clause numbers

Usual weak point

A leave and entitlements policy with sections and a version

Usual weak point

Payroll procedures and a current staff handbook

Needs shaping

A version and date on every document

Needs shaping

A named owner and a re-index step

Needs shaping

The real first job

Most organisations have some of this in good shape and some of it scattered, out of date, or quietly contradictory across documents. Getting the corpus to where every common question maps to one current, approved, citable passage is usually the larger and more valuable piece of work — it is about how your policy is captured and governed, not about buying a tool, and it is typically bigger than the AI layer on top.

Right fit if…

A small People & Culture team fields the same routine questions week after week

Your entitlements live in approved, versioned documents you can cite a clause from

Your ten most common questions each have a single, current, written answer

You want the routine layer deflected so the team's time goes to the judgement calls

Walk away if…

Your "policy" on most questions lives in a manager's head or custom and practice

Your enterprise agreement or award coverage is out of date and nobody owns the re-index

Most queries resolve to "it depends, ask me" rather than a written rule

You want a tool that gives staff binding advice without a person behind it

Open questions

The worried-buyer questions, answered straight

It only answers from the documents you have approved and loaded, and it shows the exact clause it relied on — for example a leave policy section or an enterprise-agreement clause — so the asker and People & Culture can both check it. When a question isn’t covered it refuses and escalates rather than guessing, which is the opposite of the failure mode that causes disputes under the Fair Work Act. It is a first-line reader of your own policy, not a source of new advice.

This is the most common real-world problem and the tool surfaces it rather than hiding it — if two documents address the same question differently, both passages come back cited and the answer flags the conflict instead of silently picking one. That visibility is useful, but it does not resolve the conflict; a person in People & Culture decides which instrument governs and the corpus is corrected. Reconciling those documents is usually part of the setup work.

No. It deflects the high-volume, low-variation questions — leave accrual, payday, notice periods — that fill the inbox every Monday, so the team’s time goes to the grievances, edge cases and judgement calls the tool deliberately escalates. Capacity is recaptured and redirected, not removed. Every refusal and every “needs work” flag lands with a person who owns the answer.

Answers are only as current as the documents in the corpus, so currency is the whole game. Each document carries a version and date, and when you ratify a new enterprise agreement, reissue a policy, or absorb a change like the 12% super guarantee rate or STP Phase 2 payroll reporting, the corpus has to be re-indexed for it to take effect. A stale corpus is the main way this gives a confidently outdated answer, so a defined re-index step on every policy change is part of running it.

The corpus is your own documents in a private retrieval index, and questions are answered against that index — nothing is used to train a public model. Staff questions can touch personal circumstances — a carer’s situation, a pay query — so the conversation log is treated as HR data with the same access controls and retention rules as the rest of your People & Culture records, consistent with your obligations under the Privacy Act. Who can see the logs and how long they’re kept is a decision you set, not us.

Two things. It retrieves before it answers, and the answer is built from the matched passages with the sources shown and a grounding-confidence score, so a claim with no source behind it doesn’t get made. And the refusal path is a first-class behaviour, not an error — a question like salary-sacrifice into cryptocurrency, where your policies cover super and novated leases but not crypto, returns an amber “not in your approved documents” card and an escalation, because guessing on a payroll matter is exactly what it must not do.

What it takes to build

2–4 weeks · 4 phases

Reused from template~70%

Bespoke to this skin~30%

stack · Claude · private RAG · Teams/Slack

What it would cost

Fixed scope, fixed price, fixed dates.

Bite-sized first piece

One contained change, low risk

Pilot build

Most builds land here

Embedded support

Scale on proof

Considering this for your team?

The honest place to start is a bite-sized first piece — one contained change, low risk. Tell us where it hurts; we'll play it back, scope it, and show you what's possible.

Book a call How we work