Your ServiceNow AI is moving to Amazon Bedrock. Datum CMDB already runs there.
Now Assist can now run on Amazon Bedrock models — ServiceNow-managed Claude, or models in your own AWS account. Either way, the model reads your CMDB. Datum CMDB runs on the same AI foundation and tells you, in about 30 minutes, whether that data is ready to be read.
Amazon Bedrock · Anthropic Claude · Read-only · Zero CMDB writes
A small, fast Claude model classifies and routes each check. A large Claude model reasons over the findings that need it. Every finding carries its evidence chain and a confidence score, whichever model produced it.
Your CMDB is a map of everything you run. The AI that reads it has to be trusted too.
Every server, application, owner, dependency and integration you operate is in the CMDB. Before you let an AI model read it, you should know where that model runs, who can see the data, and what it costs to do properly.
Your data does not train anyone's model
Per AWS, Amazon Bedrock does not store or use prompts and responses to train models, and model providers have no access to them. Data is encrypted in transit and at rest. Datum CMDB reads read-only and destroys assessment data on request.
Routing is why the diagnostic can be free
Inference, not consulting hours, is the marginal cost of an assessment. Bedrock lets Datum CMDB send each task to the smallest model that can do it — cutting inference cost 60–80% — so qualified organizations get the full diagnostic at no cost.
Better models without re-platforming
Bedrock puts Claude, Amazon Nova, Llama and Mistral behind one API. As stronger models ship, Datum CMDB adopts them without re-architecting — and the evidence chain behind every finding stays the same.
Now Assist on Bedrock is only as good as the CMDB it reads.
ServiceNow customers can run Now Assist skills on Amazon Bedrock two ways: ServiceNow-managed Claude, enabled from the Now Assist Admin console, or any Bedrock model in their own AWS account through the Amazon Bedrock Spoke.
Both paths hand the model the same records: the incident, the CIs behind it, their owners and their relationships. When 80% of CIs have no owner and integrations bypass reconciliation, the model summarizes the wrong server, routes to nobody, and writes a change risk summary against a dependency map that is not true.
Datum CMDB measures that before you switch it on.
Module readiness from a representative enterprise assessment of 577,000 CIs. Your scores will differ.
ServiceNow in. Amazon Bedrock in the middle. Evidence out.
Connect
Install from the ServiceNow Store. Datum CMDB reads your instance over the REST and Table APIs through a read-only MCP connection — fewer than 10,000 paginated calls, kept well under semaphore capacity.
- Zero writes to the CMDB
- Minimal payloads per call
- 40+ system tables
Reason on Bedrock
Agents on Amazon Bedrock work through six dimensions. A small Claude model routes and classifies; a large Claude model handles the analysis that needs it. Observed fact and inference are labeled separately.
- Model routing per task
- Confidence score per finding
- No training on your data
Deliver
A health score, dollar impact, evidence-based findings, a module readiness map and a phased remediation roadmap — written so a CIO and a platform owner can both act on it.
- Evidence chain per finding
- Readiness per module
- Data destroyed on request
Read-only by design
Every operation is a query. The app cannot modify your instance.
Amazon Bedrock
Models run on AWS infrastructure with encryption in transit and at rest.
No training
Per AWS, Bedrock does not use prompts or responses to train models.
Store-certified
Published on the ServiceNow Store after ServiceNow's certification review.
Check the data your AI is about to read.
Thirty minutes, read-only, zero writes. See your real CMDB health score and which Now Assist and agentic capabilities your data can support today.
Amazon Web Services, AWS and Amazon Bedrock are trademarks of Amazon.com, Inc. or its affiliates. ServiceNow and Now Assist are trademarks of ServiceNow, Inc. Platform capabilities described reflect public ServiceNow and AWS documentation as of September 2026.