Sample applications
Real applications, running against a model hosted on your own hardware.
Each app here is built the way you would build it for production — and then run twice: once against OpenAI, once against a ServeLLM node in your own region. Same code, same request, same result shape. The only difference is where the documents go.
One line changes
ServeLLM speaks the OpenAI chat-completions protocol. Point the same SDK at a region URL, swap the key, and the application code is untouched.
The data stays put
Identity documents, bank records and patient notes are processed on hardware the organisation owns, inside the jurisdiction that regulates them.
Billing stays central
Inference runs at the edge while keys, pricing, credits and a full request log stay in one control plane.
The apps
Government · Financial inclusion
LiveAsaan Qarza
A citizen uploads a CNIC and a bank statement. A vision model reads both, applies the scheme rules and returns an eligibility decision.
Identity documents and bank records are the textbook case for data residency. The same request runs against OpenAI and against a local ServeLLM node, side by side.
Open the appGovernment · Health
PlannedRural Health Triage
A basic health unit describes symptoms in Urdu or English and receives a triage category and referral guidance.
Patient data that must not leave the district, on hardware the district owns.
Government · Procurement
PlannedTender Review
Bid documents are summarised and checked against tender conditions before a committee meeting.
Pre-award commercial confidentiality, with a full audit trail in the ServeLLM request log.