Back-office automation
The workflows around the documents and the calls: reconciliation, data entry, ticket triage and reporting, across the systems you already run.
Section 1 — back-office automation
Accounts payable, employee onboarding, reconciliations.
What it does
Most back-office work is a sequence of lookups, comparisons and entries across two or three systems, with judgement needed at a few points. We automate the sequence with rules where rules suffice and models where they do not, and put a person in the loop where the rules end.
We connect over the interfaces your systems already have. No new platform sits between you and them, and every step is logged so that a reviewer or an auditor can see what happened and who decided it.
The process owner on your side stays the process owner. They see the exception queue, the throughput and the monthly report under your name.
Processes we have scoped this line for
- Accounts payable: three-way match, exception routing, posting
- Reconciliations: statement lines matched against ledger entries, discrepancy flags
- Ticket triage: category, priority and routing, with a draft reply for agent approval
- Employee onboarding: document collection, checks and system entries
- Denial triage and payment posting for healthcare revenue cycle work
- Reporting: the figures your client asks for each month, assembled from the systems that hold them
Section 2 — what it does
We do not sell licences. We build, operate and transfer. If you want a licence, we can name vendors.
How it runs
| Step | Stage | Who | What happens |
|---|---|---|---|
| 1 | Map | Person | Your process owner and our engineer write the steps, the systems and the decision points down. |
| 2 | Rules | Machine | Every step that a rule can decide is decided by a rule, and the rule is readable. |
| 3 | Models | Machine | Classification and extraction where a rule cannot decide, each with a confidence figure. |
| 4 | Exception queue | Person | Cases the rules and models cannot settle go to your team with the context attached. |
| 5 | Entry | System | Results written to your systems over the interfaces they already have. |
| 6 | Report | System | Throughput, exception rate and cycle time in the monthly report. |
Section 3 — how it runs
Steps 2 and 3 are the machine. Steps 1 and 4 are people. Steps 5 and 6 are system boundaries.
What the pilot builds
One process, two weeks, a fixed fee. Acceptance criteria are agreed before day one. Your reviewers work the queue. The report, the system and the code transfer to you.
| Process | What the pilot builds | What acceptance measures |
|---|---|---|
| Ticket classification and routing | Category, priority and routing prediction; draft reply for agent approval; review queue | Routing accuracy, draft acceptance rate by agents |
| Reconciliation | Matching of statement lines against ledger entries; discrepancy flags; exception queue | Match rate, exception rate |
| Denial triage or payment posting | Extraction from remittance advice; denial categorisation; work-queue routing; review queue | Category accuracy, posting accuracy |
Section 4 — the pilot
Targets for every measure are set in the pilot scope from your own sample. placeholder — replace with the agreed floor.
The pilot fee is fixed and quoted in the scope. placeholder — replace with the pricing model.
What you receive
- The workflow definitions, rules and integrations, documented
- The exception queue and the audit log
- A runbook your operations team can follow without us
- Transfer of all of it at acceptance
- We never contact the BPO's client.
- The BPO sets the price.
- Everything we build transfers to the BPO.
Section 5 — what you receive
Everything we build transfers to the BPO.
We do not contact your clients. Ever.