AskVault SLA breakdown per plan
Uptime target
| Plan | Monthly uptime target | Approx. allowed downtime |
|---|---|---|
| Free | None | Best effort |
| Starter | 99.5% | About 3 hr 36 min |
| Growth | 99.5% | About 3 hr 36 min |
| Business | 99.5% | About 3 hr 36 min |
| Enterprise | Custom (negotiated in your contract) | Per contract |
Uptime is measured per calendar month as (total minutes in month − downtime minutes) / total minutes in month × 100. "Downtime" means a full outage of the production API or dashboard — requests return 5xx errors or fail to connect — sustained for at least 5 consecutive minutes.
What this target is, and is not
99.5% is the availability we design, operate and alert for. It is not a contractual guarantee on self-serve plans, and we want to be exact about the difference:
- There is no service-credit scheme. We do not offer a fee credit when a month falls short.
- We do not currently publish measured uptime. We have no public status page and no third-party availability report. Publishing a percentage we cannot independently evidence would be worse than publishing none.
- We will tell you what happened. If an outage affects your workspace, email
support@askvault.coand we will explain the cause and what changed as a result.
If a contractual uptime commitment with a financial remedy is a requirement for you, that is an Enterprise conversation — see below. We would rather negotiate one we can evidence than publish one we cannot.
What counts as downtime
Downtime is defined as the inability to access core AskVault services:
- Inability to send a chat query through any channel.
- Inability to log in to the dashboard.
- Inability to make API requests to
/v1/*endpoints.
Sustained 5xx errors or failed connections on any of these for at least 5 consecutive minutes counts as downtime. Shorter blips don't qualify.
What doesn't count
The SLA excludes:
- Scheduled maintenance. Announced at least 24 hours in advance by email, capped at 4 hours per calendar quarter.
- Customer-caused outages. Misconfigured webhooks, bad credentials, exhausted query quota, a revoked API key, or breach of the Terms of Service.
- Force majeure. Natural disasters, war, regional internet outages, government action, or other events beyond our reasonable control.
- DDoS attacks targeting your specific workspace. We mitigate at the edge, but DDoS-related slowness on your specific endpoint isn't general AskVault downtime.
- Issues with sub-processor LLM providers. We design our generation pipeline for resilience to individual provider issues, but extended upstream outages may briefly affect answer latency.
- Beta or experimental features explicitly labelled as such in the dashboard or release notes.
Support response time
AskVault does not publish per-plan support response-time SLAs, and does not currently commit to a response time on self-serve plans — we are a small team and would rather not promise a number we do not measure. In practice you are usually replying to the people who build the product. Enterprise customers can negotiate a dedicated support response commitment as part of their contract.
Incident acknowledgement and communication
We notify workspace owners by email about unplanned outages that affect them, and follow up with the cause and what changed as a result. We do not currently commit to an acknowledgement or follow-up window on self-serve plans, because we have no on-call rota to hold it up. Enterprise contracts can define one.
Enterprise SLA customization
Enterprise contracts can customize:
- Higher uptime target, negotiated in your Master Services Agreement.
- Service credits. Self-serve plans carry none; an Enterprise agreement can define a credit or remedy schedule.
- Faster acknowledgement window, a dedicated support contact, and on-call escalation.
- Maintenance window exclusions. No maintenance during your business hours.
Contact sales@askvault.co for SLA discussion. Each Enterprise SLA is contract-specific.
Calculating your effective uptime
To compute your workspace's effective uptime over a month:
- Total minutes in month (about 43,200 for a 30-day month).
- Subtract minutes of downtime, based on your own records. We do not currently publish measured uptime, so this is your figure rather than a reconciled one.
- Divide.
For example, 30 minutes of downtime in a 30-day month: (43,200 - 30) / 43,200 = 99.93%. Above 99.5% means the target is met; below means a credit applies per the ladder above.
FAQ
Does the SLA apply to channels like WhatsApp or Telegram?
Yes for AskVault's own pipeline. We commit to the SLA for our processing of WhatsApp messages. We don't control WhatsApp's own uptime; if Meta has an outage, that's outside our SLA.
What about LLM provider outages?
We design our generation pipeline for resilience to individual provider issues, so most upstream LLM outages are handled transparently. Extended outages affecting our upstream providers are rare; when they happen, we still count them against our SLA, because maintaining sufficient resilience is our responsibility.
Can I get a refund instead of credit?
Credits are the default remedy. Enterprise contracts can negotiate refund terms. Reach out to sales@askvault.co.
Does the SLA apply during my free trial?
No. The Free plan, including its 14-day trial, is best-effort with no SLA. SLA coverage starts once your workspace moves to a paid plan.
How do I know if my workspace is healthy?
Two sources: your dashboard's Channel Health view, and the analytics for unanswered queries or error rates.