AWS shows synthetic monitoring with Nova Act and AgentCore
AWS published a tutorial on building synthetic monitoring using Amazon Nova Act and Amazon Bedrock AgentCore. The approach replaces traditional selector-based browser scripts with natural-language actions driven by Nova Act's multimodal model, which processes UI screenshots instead of DOM locators. AWS says early enterprise customers have seen over 90% accuracy on browser workflows with Nova Act.
Key points
- Nova Act uses vision-language model on screenshots, not DOM selectors, for resilience
- Architecture combines EventBridge Scheduler, AgentCore Runtime, Browser tool, and SNS
- Sample ecommerce journey runs in 2-4 minutes; ~8,640 runs per month at 5-minute cadence
The architecture uses EventBridge Scheduler to trigger runs every 5 minutes to hourly, AgentCore Runtime for serverless execution with session isolation, the AgentCore Browser tool for remote browser sessions, and SNS for failure alerts. A sample repository provides a complete implementation for a six-step ecommerce journey that typically completes in 2 to 4 minutes. AWS includes a deploy.py script and a CDK stack for production deployment with dead-letter queues and CloudWatch alarms.
Cost estimates for the sample six-step journey running every 5 minutes (about 8,640 invocations per month) cover AgentCore Runtime invocations, Browser tool sessions, Nova Act inference, EventBridge Scheduler calls, and SNS notifications. AWS recommends focusing on 3-5 critical workflows to avoid alert fatigue and using ephemeral browser sessions for test independence.
Implementing synthetic monitoring using Amazon Nova Act
AWS Machine Learning Blog · 28 September 2026
Loading the full article…
This text was published by AWS Machine Learning Blog and written by Sarath Krishnan. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Agents & Tools
All →- Modulate raises $25M to analyze voice tone and intent beyond transcripts · 2 src
- Aws releases vLLM-Omni dlc for image and video generation on SageMaker AI · 2 src
- Meta Muse agent allegedly sold item and shared address without permission · 2 src
- TypeSafe AI launches Jev, a decision-focused AI 193x faster and 444x cheaper than ChatGPT · 3 src
- OpenAI says AI agents leaked 53 images from ChatGPT users · 9 src
Comments
via GitHub Discussions