Frequently Asked

It depends entirely on the site. A deployment starts with a visit, not a fixed timeline, because the right hardware and the right training data depend on what is actually running at that site already. A straightforward vision deployment on cameras already in place moves faster than a knowledge system built from years of scattered paper records.

Yes. A deployment can run fully offline where the client requires it. Where a client wants an optional integration with another system or software product, that connection is scoped and agreed separately, not switched on by default.

The client. Hardware is specified or supplied as part of a deployment, but ownership sits with the organization it is installed for, the same as any other equipment on their site.

Most organizations' data looks like this, which is exactly why it sits unread in the first place. A site visit and data review happen before training starts, specifically to work out what is usable and what needs cleaning up first. This is normal, not a blocker.

Yes. A single camera feed, one archive of documents, or one production line is enough to start with. The four-step process, train, deploy, adapt, compound, works the same way at small scale as it does at large scale.

The trained model belongs to the client once it is installed, along with the hardware it runs on. Nothing about the arrangement locks an organization into an ongoing dependency they cannot leave.

Yes. Knowledge models built on the Sijill family read and answer in Urdu, English, or a mix of both, depending on how the client's own records are written.

A reply comes within two working days. That first conversation covers what the client already runs and what sits unread, before anything is proposed.

Pricing is scoped per deployment, based on the hardware a specific model needs and the complexity of the training involved. There is no fixed price list, because a document search system and a sixteen camera vision system are not the same job.