Systems that have to survive an audit.
We build software for industries where being wrong is expensive and someone will eventually ask you to prove what happened — hospitals in India, and certified manufacturers in the Gulf.
How we build
The same three ideas turn up in everything we build.
We reached each of them separately, on unrelated projects, because the same problem kept coming back: the records a system produces have to hold up when someone comes to inspect them.
01
A record that cannot be quietly edited
Every action is written to a log that can be added to but never changed, and each entry is sealed against the one before it. Alter an old record and the chain breaks. A check re-walks it and tells you exactly where.
02
Permission that starts at no
Consent, approval and whether a document is still in force are each tracked in their own right, and each one starts as a no. If the system cannot confirm that permission exists, the action stops rather than going ahead.
03
AI that cannot sign anything
Models draft documents, suggest matches and analyse what is there. Only a person approves. That limit is built into the software rather than written into a policy someone could set aside.
Products
Two products, two rulebooks.
Synaptica
India · hospitals
A hospital information system built for Indian rules from the start. Patient consent and the handling of personal data sit in the foundation: the system checks them before it acts, rather than having them added later to satisfy an inspection.
StatusIn build. Not yet deployed.
synaptica.in →
Doqumus
Gulf · certified manufacturers
Quality management software for manufacturers certified to ISO 9001 and API Spec Q1. The part that has to be validated is kept small and changes rarely. The AI sits outside it, where it can draft and analyse but never approve.
StatusDoqumus Works is in production. Document control is in build.
doqumus.ai →