Last updated 15 August 2026
Pilot Service Levels
Practical service expectations for pilot customers. These statements describe the current operating model and are not an enterprise contractual SLA.
Processing expectations
Claim photo analysis is asynchronous. Accepted submissions return a session identifier immediately, then processing continues in the background until the result endpoint reports a terminal status or a review state.
- For standard pilot batches of up to 10 photos, customers should normally expect results in minutes when service capacity and required dependencies are healthy.
- An individual long-running analysis attempt is capped at 15 minutes by the current service configuration. Large files, high image counts, service contention, or dependency delays can extend the elapsed time before a final status is available.
- Customers should treat webhooks as notifications and fetch the authenticated result endpoint for the canonical status before making claim decisions.
Webhook delivery
Webhook payloads are signed with HMAC-SHA256. Customer systems should verify the signature, store the delivery identifier for idempotency, and return a 2xx response only after accepting the event.
- Automatic webhook delivery currently makes 3 total attempts.
- Automatic retry delays are 30 seconds then 2 minutes after failed attempts.
- Delivery attempts are recorded with response status, response body snippet, timestamps, attempt count, and next retry metadata for support diagnosis.
- Manual replay may be used by authorised users when a customer endpoint was unavailable or returned an unexpected response.
Retention and access
Pilot retention is configuration-led and must be agreed with each customer before automated deletion is enabled.
- Evidence sessions and uploaded files use a default pruning window of 60 days when evidence pruning is enabled.
- Access audit logs use a default pruning window of 365 days when audit pruning is enabled.
- Actual retention may be longer where claim-handling, dispute, regulatory, legal, or customer contract requirements apply.
See the Retention Summary for more detail.
Support and incidents
- Operational support is available through support@claimsight.io.
- If a material incident affects processing, webhook delivery, result access, reports, or customer data confidentiality, Claimsight will contact affected pilot customers through the agreed support or security contact route as soon as practicable.
- The standard pilot service does not promise 24/7 monitoring or a 24-hour notification target. Any monitored hours, escalation route or additional target must be operationally confirmed and recorded in the customer agreement.
- Security reports should be sent to security@claimsight.io.
Planned maintenance
Planned maintenance may be scheduled during lower-volume periods. Where maintenance is expected to interrupt uploads, result retrieval, report downloads, or webhook delivery, pilot customers will be notified through the agreed operational contact route.