Dynatrace used Perform 2026 to announce several updates that directly affect how developers work. The changes focus on reducing debugging time, adding AI-powered automation, and connecting frontend and backend observability in ways that weren't possible before.
What Changed
The company introduced Dynatrace Intelligence, which combines two types of AI. The first is deterministic AI - it uses real-time data about how your systems actually behave to find problems. The second is agentic AI - it can reason through issues and take action automatically within the guardrails you set.
This matters because it addresses a real problem. When you deploy AI-powered features or agentic workflows, they behave unpredictably. Traditional monitoring tools show you metrics, but they can't tell you why an LLM response took 8 seconds instead of 2, or why a specific user session failed.
Andreas Grabner, Fellow of Developer Advocate at Dynatrace, explains: "As organizations adopt AI at scale, as they start to build AI-native services, understanding how those systems behave in real environments becomes increasingly important. When frontend, backend, cloud, workflow and AI telemetry are connected through a single causal view, teams can see how user experience, application behavior, infrastructure, AI models and agentic workflows influence one another."
Real User Monitoring Gets an Upgrade
The new RUM capabilities address specific problems with modern web apps. Single-page applications, soft navigations, and asynchronous rendering create blind spots in traditional RUM tools. You see the metrics, but you can't connect frontend errors to backend traces or understand why users abandoned a session.
Dynatrace now stores RUM data in Grail, their data lakehouse, alongside logs, metrics, and traces. This means you can query frontend performance in the context of everything else. The Error Inspector app groups related errors and shows you the full context - not just the stack trace, but the user journey, backend services involved, and infrastructure state.
Dana Harrison at TELUS says developers there spend significantly less time debugging thanks to these tools. The company reduced deployment time for end-to-end observability from 600 minutes to 20 minutes using automation and monitoring as code.
IDE Integration and Live Debugging
Dynatrace expanded IDE support to include Windsurf and Cursor. Live Debugger now works directly in your programming environment, so you don't need to context-switch between your code and a separate observability platform.
This also includes mobile debugging. The new ANR (Application Not Responding) and crash diagnostics give you the context you need to fix mobile issues faster. Instead of generic crash reports, you get symbolized stack traces with full observability context.
Multi-Cloud Integrations
The platform now has deeper integrations with AWS, Azure, and Google Cloud. This matters if you're running services across multiple clouds or using cloud-native services like Azure AI Foundry or AWS Bedrock.
The integrations surface health indicators and automated alerts for cloud-native workloads. More importantly, they connect cloud infrastructure state to application behavior. When something goes wrong, you can see if it's your code, the cloud service, or how they're interacting.
Agentic Workflows and MCP Support
Dynatrace added support for the Model Context Protocol (MCP) and built agentic workflows into the platform. This means AI agents - whether from Claude, AWS Bedrock AgentCore, or Azure AI Foundry - can safely act on observability data.
The company also announced domain-specific agents for SRE, development, and security teams. These agents can triage issues, generate remediation plans, and coordinate with external systems like ServiceNow or GitHub.
When Dynatrace benchmarked an external SRE agent working with their deterministic agents, problems were solved up to 12 times more often, three times faster, and at half the cost compared to tests without deterministic agents.
DevCycle Acquisition
Dynatrace acquired DevCycle, a feature flag and experimentation platform. This adds runtime control at the feature level. You can validate behavior continuously and automatically respond to issues without redeploying.
The integration work is underway, with more updates coming as the acquisition progresses.
What This Means for Your Work
These changes address three common pain points:
Debugging AI features: You can now trace LLM calls, token usage, and model behavior alongside application code.
Frontend-backend correlation: No more guessing which backend service caused a frontend error. The full trace is connected.
Reducing manual work: Agents handle repeatable tasks like incident triage, vulnerability assessment, and performance optimization.
The platform aims to turn observability from a passive system that shows you problems into an active system that helps you fix them. Whether that saves you time depends on your specific workflow, but the capabilities are there.
Availability
Most features are available now. Azure and GCP integrations are in preview. DevCycle integration details will come later as the acquisition work continues.
If you're already using Dynatrace, these updates roll out through the normal update process. If you're evaluating observability platforms, the company offers a 15-day trial.
