Waterside AI

WorkNDIS allied health provider, Sydney

The manual document check is gone. The manager acts on exceptions only.

A workflow audit with the practice owner, then a compliance system that checks practitioner documents against the capacity register, sends reminders and payment-withheld notices on its own, and lists who is overdue each fortnight.

A treatment room with an adjustable table, ready for the next appointment
Business
NDIS allied health provider, Sydney
Scope
Workflow audit, process mapping, compliance automation, AI agent specification
Status
Built and deployed. AI agent designed and specified.
Works with
Notion, Gmail, n8n

What happens on its own now

  • Document goes overdueFlagged against the capacity register
  • Payment status changesReminder or payment-withheld notice sent
  • Fortnight endsSummary lists who is overdue, on which document

The business

An NDIS allied health provider in Sydney with ten practitioners, each required to hold a current set of compliance documents. The practice has to know, at any moment, who is compliant, who is overdue and on which document.

The problem

Booking, scheduling, invoicing and reporting each ran as a separate manual process. Practitioner documents were checked by hand. A reminder workflow already existed and kept failing: its drafts landed in an account the practice could not see, its duplicate check had been switched off so the same practitioner was chased twice for the same document, and drafts were lost without being actioned.

What was built

  • Workflow audit

    Discovery with the practice owner. Current and future processes mapped for booking, scheduling, invoicing and reporting, and the business rules for compliance defined with the owner in plain language.

  • Compliance system

    Checks practitioner documents against the capacity register under the agreed rules. Overdue documents are flagged. Reminders and payment-withheld notices go out automatically based on payment status. A fortnightly summary lists exactly who is overdue and on which document.

  • AI agent with guardrailsDesigned and specified

    Diagnosed why the existing reminder workflow kept failing and designed an AI agent to replace it, tracking 220 compliance documents across 10 practitioners in one clear email per practitioner. Fixed rules read the data and the model decides only escalation and wording, which keeps cost predictable and behaviour auditable. Three independent guardrails mean it can only ever email the practice manager. Tests for dry runs, duplicate sends and error handling were specified before deployment. This agent is a specification: it has not been built or deployed.

What it does for the owner now

  • The manual document check is gone.
  • The manager acts on exceptions only. Overdue documents are flagged, and the fortnightly summary says who and which.
  • Reminders and payment-withheld notices go out on their own, driven by payment status.
  • A replacement for the failing reminder workflow is specified in full, with its guardrails and tests written down before any of it is built.
220compliance documents across 10 practitioners, tracked by the specified agent

Sound familiar?

If your practice still checks practitioner documents by hand, or runs a reminder workflow nobody quite trusts, this is where an engagement starts.

One email starts it.

Tell us what your business does and which part of it still runs on someone remembering. Every enquiry gets a considered reply.