Akuru's own Dhivehi models inside your systems: English drafts of announcements, searchable archives, written records of meetings, and services people can listen to.
Staff write in Dhivehi; an English draft appears automatically; a person checks it and publishes.
Scanned letters, gazettes, forms and records become text you can search, copy and translate.
Recordings of meetings, council sittings and calls become Dhivehi text, ready to edit into minutes.
Notices, forms and replies read aloud in Dhivehi, for people who find reading hard.
One API with keys per system, or a ready kit for websites on Cloudflare: your developers add it in a day.
Machine translations are drafts. Your staff approve what is published, and the drafts are labelled as machine translation until they do.
Better models reach your systems automatically. Settings are controlled centrally, and translation can be paused instantly.
Work is queued and retried; if a model is busy, the job waits instead of failing, and nothing your staff save is lost.
We set it up on one of your services, measure the results with your staff, and decide together.