Drapto has three AI agents, each doing one job well. They run on your own AI provider key, work only from data already in your account, and every figure they surface traces back to your own books.
An agent that can invent a number is worse than no agent.
Clinic finance cannot run on plausible-sounding estimates. Drapto calculates deterministically first and uses the model only to explain and prioritise what the arithmetic already found — so the numbers cannot be hallucinated.
It works through your collection gaps, unbilled procedures, write-offs and ageing balances, then tells you where the recoverable money is sitting and in what order to chase it. Every figure it reports is verifiable against your own books, because the calculation is deterministic — the model explains the result, it does not produce it.
Monthly, it reviews each clinician's activity, contribution and patient satisfaction and writes an individual view: what improved, what slipped, and what would help. It runs on data you already have, and the monthly cadence keeps the cost negligible.
It identifies patients who have gone dormant, segments them by value and how long they have lapsed, and drafts the outreach for each segment. Reaching a patient who already knows you costs a fraction of acquiring a new one.
A prioritised list of accounts beats a report telling you collections are down.
A monthly written view per clinician gives directors something to open a conversation with, rather than a ranking that starts an argument.
Most clinics have several years of lapsed patients nobody has systematically contacted.
You supply the AI key, choose the model, and see usage directly — there is no per-word agent surcharge from us.
You connect your own AI provider key and the agents run on it. You choose the model, you see the usage, and your clinic data is not pooled into anyone else's system. Keys are encrypted at rest and scoped to your account.
The financial calculations are deterministic — they are computed from your records, not generated by a language model. The model is used to explain, prioritise and write, which means every number it presents can be checked against your own books.
No. They produce analysis, coaching notes and drafts. A person reviews and approves before anything is sent or acted upon.
It is deliberately written as individual coaching rather than a public league table, because ranking clinicians on revenue is the fastest way to lose their engagement. Output focuses on what changed and what would help.
Yes, individually and permanently. Agents are optional and the platform works fully without them.