receipting.ai now covers end-to-end accounts receivable for insurance

For the past six years, we’ve been building bespoke AI for insurance finance teams. Custom extraction models trained on broker remittances, policy documents, bank statements. Each new document format meant gathering samples and retraining. Each new country or region meant a fresh training dataset. We could keep one step of the process — extraction and matching — working well, but it was a treadmill. And tackling a completely different problem, like processing closing advices or automating dunning, would have meant building another custom model from scratch.

Then general-purpose AI changed the economics. Instead of training a model for each document type, you could point the same system at a closing advice, a remittance, a broker reply, a dunning response — and it just reads them. No training data. No retraining cycle. The thing that had locked us into solving one step at a time was gone.

But the product insight didn’t come from staring at the technology. It came from talking to underwriters. When we started asking insurance finance teams what else was painful — not just “can you match this remittance faster” but “what’s actually broken in your AR process” — we heard the same answer over and over: everything is connected, and fixing one step in isolation only moves the needle so much. That’s when we stopped building a matching tool and started building an AR system.

As of April 2026, receipting.ai covers the full accounts receivable lifecycle for insurance.

What “end to end” actually means

Most AR tools do one thing. They match payments. Or they send dunning emails. Or they give you a dashboard. None of them handle the full lifecycle for insurance, where the data is messy, the documents are non-standard, and a single transaction can take months to resolve.

We built four capabilities that work together as stages in a single flow:

Closing advice processing. Brokers send closing advices 30-60 days before payment. They tell you exactly what’s coming — which policies, what amounts, any adjustments. We read them on arrival, match them to outstanding policies, and flag discrepancies before the money even lands. This is the stage that prevents suspense items from ever being created. Nobody else in the AR space automates it because nobody else seems to know it exists. (In an upcoming post is a detailed piece on why this is the highest-leverage change most finance teams aren’t making.)

Remittance matching. When payments arrive, AI reads the remittance — PDF, CSV, Excel, email body, whatever the broker sends, in whatever language — and matches each line to outstanding premiums. We’re hitting 97%+ straight-through processing from day one across live customer accounts. That climbs to 99%+ when closing advice processing is active.

Suspense line tracking. Unmatched lines don’t go into a spreadsheet to die. They go into a managed queue with two-way email. Your team can query a broker directly from the platform, and when the broker replies, AI reads the response and logs it against the correct line. I saw a CFO’s reconciliation spreadsheet last year — one tab per broker, 47 tabs, colour-coded by how overdue each item was. That’s what we’re replacing.

Automated broker chasing. Overdue premiums get chased on schedule. Every communication is logged. Your finance team handles exceptions, not the routine “where’s our money” emails that nobody enjoys writing.

Why these have to be connected

The insight that took six years to land is that these aren’t four separate products. They’re four stages of the same flow, and the value comes from the connections between them.

Closing advice processing prevents suspense. But only if the pre-matched data feeds into remittance matching so the system knows what to expect when payment arrives.

Suspense tracking resolves the items that slip through. But only if it has context from the matching stage about what was tried and why it didn’t match.

Dunning chases overdue brokers. But only if it knows which items are genuinely overdue versus already matched and pending posting.

Here’s a less obvious example: broker contact information. Brokers regularly change which person is responsible for a particular policy as staff and responsibilities shift. When the system processes closing advices and remittances, it doesn’t just extract data — it reads the context of each email to understand who the right person is to contact about a specific policy. That keeps your contact information current without anyone maintaining it. And when the chasing stage sends a follow-up, it’s going to the right person, not the person who left six months ago.

Disconnect any one of these and you’re back to optimising individual steps — which is how most finance teams have been stuck for years. Connect them and the compound effect is substantial: problems prevented, matching accelerated, exceptions resolved faster, and brokers chased before items go stale.

Why generic AR tools don’t work here

I get pitched generic AR tools regularly. They all say they handle “payment matching.” But they don’t understand policy numbers. They don’t know what a closing advice is. They can’t split a $247,000 bulk payment across 83 policy lines from different coverholders. They’ve never seen a remittance where the broker uses their own reference numbers instead of yours and spells the insured’s name three different ways.

Insurance AR has specific problems: premium trust accounts that need strict reconciliation, multi-policy remittances that need individual line matching, closing advices that are unique to this industry, and suspense management that’s a compliance concern not just an efficiency one. You need a tool built for this.

What’s next

We’re rolling this out to more insurance underwriters and MGAs across Australia, the UK, and the US. If your team is spending their best hours on matching and managing suspense in spreadsheets, I’d like to show you what we built. Book a demo — bring your worst remittance and we’ll match it live on the call.