ANVE: the private AI office, on the record. See what that means

ANVE

Private AI glossary · updated 2026-10-16

Service level agreement (SLA)
The measurable promises behind a service, and what happens when they are missed.

Short answer. A service level agreement (SLA) is a commitment between a service provider and its customers that sets specific aspects of the service, such as responsibilities, expected performance, reliability and response times, and requirements for reporting, resolution and termination. NIST’s glossary defines it in these terms. Take a cloud example: Google Workspace promises at least 99.9% monthly uptime, and if it misses, the remedy is service credits of 3 to 15 days of service, which the SLA calls the customer’s sole and exclusive remedy. Even 99.9% allows about 43 minutes of downtime in a 30-day month. For an AI box in your office, the SLA that matters is different: on-site response and repair times, spare parts and update timelines. India’s Manual for Procurement of Goods lists uptime, response time, escalation and penalties as SLA terms in maintenance contracts.

01

Reading an uptime SLA

Google Workspace’s credits, with the downtime each level allows in a 30-day month (43,200 minutes).

Monthly uptimeDowntime allowedGoogle Workspace credit
99.9% or moreUp to about 43 minutesNone: the SLA is met
Below 99.9%, down to 99.0%Up to about 7.2 hours3 days of service
Below 99.0%, down to 95.0%Up to 36 hours7 days of service
Below 95.0%More than 36 hours15 days of service
02

Fine print that matters

  • Definition of downtime: Google counts downtime when the user interface has more than a 5% user error rate, measured on its servers.
  • Claiming: the customer must request the credit, within 30 days for Google Workspace.
  • Cap: Google’s credits are capped at 15 days of service in a month.
  • Exclusions: force majeure, and problems caused by the customer’s own equipment.
  • Sole remedy: credits may be the only compensation, whatever the outage cost you.
  • Coverage: check that the AI features you rely on are covered services.
03

SLAs for on-premise AI

When the AI runs in your office, uptime depends on your power, your network and the box itself. The SLA to negotiate is about the vendor’s response: how fast it answers a fault call, how long repairs and part replacements take, how quickly security updates arrive, and what happens to the data on a failed drive. The procurement manual suggests terms such as percentage uptime, channels for service requests, response times, escalation, monitoring and penalties for delay.

As the /security page sets out, ANVE does not update itself over the internet: today ANVEAI installs updates on site, and signed update packages applied offline by USB are not on ANVE yet. ANVE is pre-release, with no delivery date and no final price, so get response and repair times in writing before a paid pilot.

04

Questions people ask

What does 99.9% uptime mean?
That the service is available at least 99.9% of the time in a month, which still allows about 43 minutes of downtime in a 30-day month.
Are service credits the same as compensation?
No. They are usually extra days of service or a credit on a future bill, capped, and often the only remedy the SLA allows.
What SLA should we ask an on-premise AI vendor for?
Fault response and repair times, part replacement, timelines for security updates and named escalation contacts, with penalties for delay, as the procurement manual suggests for maintenance contracts.
05

What is not real yet

No sealed ANVE unit exists; every image of the appliance on this site is a labelled render. The software runs today on a development machine. No customers yet, no certifications, no delivery date, and the final price is not announced.

Reservations are ₹999 in India or US$12 elsewhere, refundable in full until you confirm an order in writing (terms). Offices that cannot use cloud AI can apply to be one of ten design partners.