Two ways to run it.
Same wire, same file, same gate, same operator. What differs is whose machine it runs on.
ATAILA Enterprise
operated Your hardware, in your building, yours alone: one tenant on one machine. The file never leaves your building. ATAILA installs it, operates it and keeps it updated. Two cases justify it: you already own GPUs, so this is a second workload on capacity that is idle most of the day; or your workload mix justifies the machine on its own.
Quoted per case. The packaged release with an update manager is on the roadmap building .
ATAILA Cloud
operated ATAILA's operated EU platform. The hardware is shared with other workloads on machines ATAILA owns and runs; your file is a separate index on it. No capital expense, one contract, one accountable party, an EU jurisdiction. You give up "the machine is physically mine". You keep everything the product is about: the beat, the file, the spike log, the record.
For most companies this is the right place to start.
What ATAILA operates, in every case
The wire: the beat configured, the sweeps run, the candidates queued with source, date and publisher. The drafts, from your file, in both languages. The capture of every source you answer, hashed and dated. The platform underneath: patches, backups, updates. The one thing ATAILA never does for you is the approval. Your named person releases what goes out.
If you stop
Your file, your spike log, your archive and your published content are yours. Export is a contract term, not a favour: the file and the spike log as files, the archive with its manifests and hashes, the pieces as you published them.
Do not buy a GPU for this
- Start on ATAILA Cloud. For most companies that is the answer.
- You already own GPUs. Then this is a second workload on idle capacity, and the file stays home.
- Buy the machine when the workload mix justifies it. The newsroom, a helpdesk, contract analysis and a coding assistant, together.
A GPU bought for one marketing workload sits idle most of the day.