Scope & Billing Guardian
Reviews each draft invoice against the engagement letter scope and agreed rate card before it goes out — flagging out-of-scope work, rate variances and non-billable time, and drafting the adjustment note for the engagement owner to clear.
It cross-checks every line on a pre-bill against the engagement scope and rate card, then drafts the exceptions for a person to approve before the invoice is sent.
Most mid-sized professional services firms lose a slice of revenue not to bad work but to the gap between what was scoped and what was billed: out-of-scope work absorbed without a variation, time logged at the wrong rate, non-billable hours that slip onto the invoice. By the time anyone notices, the draft has gone out and the conversation is awkward. Scope & Billing Guardian reads each pre-bill before it leaves the building, checks every line against the engagement letter and rate card, and hands the engagement owner the exceptions — and a drafted note — to clear.
Where this works well
The slow, invisible problem this makes visible is the line-by-line reconciliation that should happen on every draft invoice and almost never does. A billing manager is meant to check each time entry against the engagement letter scope, the agreed rate card, and the non-billable rules — and across dozens of drafts a month, under deadline, that check gets skimmed. So scope creep (a Phase 4 workshop billed under a fixed fee that only covers Phases 1–3), rate variances ($480/hr charged where the engagement says $420), and non-billable travel land on invoices and either go out wrong or get caught late and written off.
It earns its keep for the engagement owner or billing manager at a firm running enough concurrent engagements that no one can hold every scope and rate in their head — typically a mid-sized advisory, accounting, engineering or law practice on a mix of fixed-fee and time-and-materials work. It is most useful where there is a real, written engagement letter and rate card to check against, and where the cost of an over-bill (a client dispute, a write-off, an awkward variation conversation) is high enough to be worth catching before the invoice is sent.
Where it works badly
It is only as good as the engagement terms it checks against. If scope lives in a partner's recollection rather than a written engagement letter, the tool has nothing to compare a line to, and it will either flag everything or flag nothing. If a variation was agreed and signed but never recorded in the system, the tool will confidently flag legitimately-billable work as out of scope — a wrong result delivered with the same certainty as a right one. The honest test: pull your last ten draft invoices and try to point, for each questionable line, at the clause that governs it. If you can't find the clause quickly, neither can the tool.
It is also a poor fit for firms whose billing is genuinely judgement-heavy and lightly documented — value-based pricing set per matter, blended rates negotiated verbally, "we'll sort it out at the end" engagements. There the exception list would be noise, because almost every line is an exception to a rule that was never written down. Fixing that is a billing-discipline problem before it is an AI problem.
What it doesn't do — and shouldn't
It surfaces; the engagement owner decides. The tool flags that Phase 4 work falls outside the fixed-fee scope and recommends a signed variation — it does not raise the variation, adjust the invoice, or send anything. It flags a rate variance — it does not decide whether to honour the higher rate as a goodwill gesture or correct it. It marks a draft "held for review"; it never holds a client relationship.
That boundary is deliberate, and in professional services it matters more than usual. Whether to bill out-of-scope work, absorb it, or raise a variation is a commercial and relationship judgement that can affect a client relationship worth far more than the line in question — and for law practices, what is disclosed and charged sits under costs-disclosure obligations that a person, not a workflow, must own. The tool gets the firm to the decision faster with the evidence attached. The decision stays with the human.
What your data has to look like for this to work
This is the part most firms under-estimate, and it is the part we actually help with. For the tool to check a line, three things have to exist in a form it can read. First, the engagement scope: which phases, deliverables or work types the fee covers, tied to the engagement — not just a fee number, but what the fee is *for*. Second, the agreed rate card per role (Partner, Senior Consultant, Analyst, and so on), and which engagement each rate applies to. Third, the non-billable rules — travel, internal time, write-ups — stated explicitly rather than assumed. And the time entries themselves have to carry enough structure (which engagement, which phase or activity, which timekeeper) to be matched against all three.
Most firms have some of this in good shape and some of it scattered across PDFs, email threads and a partner's memory. Getting the scope, rate card and billable rules into a consistent, machine-readable shape — usually by changing how the engagement letter and time entry are captured, not by buying a new tool — is typically the real first job, and it is bigger and more valuable than the AI layer on top of it. That groundwork is what makes every later check trustworthy, and it is the work we scope with you first.
Could it flag the wrong thing and hold up a clean invoice — or wave through one it shouldn't?
It never sends or holds an invoice on its own. Every draft it reviews is presented to the engagement owner with the exceptions it found, each pinned to the clause it relied on (engagement letter §1.2, rate card §4.1), so you can see its reasoning and overrule it in one click. A draft it marks "Clean" is still yours to send or not; a draft it flags is held only in the sense that it asks you to look before it goes out. The safeguard is that the human approves the bill, not the tool.
Our engagement letters and rate cards are inconsistent and half of them are in PDFs — will this even work?
Probably not well on day one, and that is the honest answer. The tool can only check a line against scope and rate if the engagement scope (which phases or deliverables the fee covers), the agreed rate per role, and the non-billable rules are written down somewhere it can read them. If your engagements are priced in email threads or partners' heads, getting that into a consistent, machine-readable form is the real first job — and it is usually more valuable than the AI layer itself.
Does this replace our billing manager or the partner who reviews WIP?
No. It replaces the line-by-line cross-checking that a billing manager does by eye against the engagement letter — the slow, easy-to-skip part — so the person who reviews the bill spends their time on the judgement calls the tool surfaces. Whether Phase 4 gets billed, whether to raise a variation, whether to write off the travel as a relationship gesture — those stay with the partner or engagement owner. Capacity is recaptured for that judgement, not removed.
How current does the engagement and time data have to be?
Current enough that the rate card and scope it checks against are the ones in force for that engagement, and that the time entries are the ones actually about to be billed. If a variation was signed last week but not recorded, the tool will flag legitimately-billable Phase 4 work as out of scope — a confident-but-wrong result. It reads what the system holds at review time, so stale scope or an out-of-date rate card produces stale flags.
Where does our client and billing data go — does it leave our systems?
The pattern runs against your accounting and practice-management systems (for example Xero or your practice system) through their APIs, and the review and approval happen in your workflow (e.g. Teams). What the LLM sees and where it runs is a deployment decision we scope with you up front, with Australian data-handling and your client-confidentiality obligations as the frame. Nothing is sent to a client without a person approving it.
Estimated build: 3–5 weeks. Most of it is template work we've already done.
Fixed scope, fixed price, fixed dates.
The cost band reflects the engagement shape, not a per-feature line item. We work on fixed scope, fixed price, fixed dates — see the services catalogue for what falls inside each band.
Considering this for your org?
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.