Dynatrace makes observability an active control layer for product teams
At its Perform conference in Las Vegas, Dynatrace announced a set of developer experience and agentic AI updates aimed at one goal: move from passive monitoring to real-time control of software behavior in production. The pitch is straightforward-give teams the context to act fast, the levers to change behavior safely, and the automation to keep systems steady without waiting on redeploys.
The shift is timely. Dynatrace research indicates product development is the fastest-growing use case for agentic AI: 42% of organizations already use it in customer-facing products, and another 31% plan to expand within five years. Releases are getting faster, and systems are more dynamic. Visibility is table stakes; control is the differentiator.
What's new
- Unified frontend observability with RUM in Grail: A refreshed frontend experience and new apps (including Error Inspector) give teams deeper visibility into real user behavior and faster issue isolation.
- Expanded mobile diagnostics: Improved context for ANRs and crashes, making it easier to pinpoint the root cause and cut debugging time.
- Runtime controls via DevCycle acquisition: Feature-level controls combined with Dynatrace Intelligence to validate behavior continuously, reduce risk, and auto-respond from development through production. Integration is underway, with additional updates to follow.
- End-to-end traces across AI, databases, and cloud: Unified tracing links AI calls, app services, databases, and infrastructure for clarity as AI-driven workloads increase trace volume and complexity.
- Agentic workflows and MCP Server: Safe, real-time actions for developers and AI agents using observability data. MCP support includes Claude, AWS Bedrock AgentCore, and Azure AI Foundry for multi-cloud and multi-AI environments.
- Live Debugger in more IDEs: Expanded support now includes Windsurf and Cursor, bringing instant, in-IDE access to live debugging.
Why this matters for product development
Release risk often hides in the last mile: feature flags misconfigured, prompts drifting, or a dependency governed by someone else's change window. Dynatrace is pushing toward a model where those risks are managed at runtime with precise controls and immediate feedback loops. That means fewer coordination gaps and faster recovery when behavior drifts.
For teams shipping AI-assisted and agentic features, tracing across AI calls, backend services, and cloud resources is crucial. The ability to link user behavior (RUM), mobile ANRs, and distributed traces in one workflow shortens the loop from incident to fix. And with agentic workflows, you can automate safe actions that respect guardrails.
How to put this to work this quarter
- Instrument key user journeys with RUM and wire Error Inspector alerts to the teams that own those flows.
- Map your top mobile ANR/crash offenders and fix the top 3 by volume and user impact first.
- Stand up feature-level runtime controls (via DevCycle integration) for high-risk releases; require canary by default.
- Trace AI calls end-to-end and set SLOs on latency, cost, and failure modes (timeouts, 429s, content filters).
- Enable Live Debugger in the IDEs your team uses most and codify a "first responder" routine for production issues.
- Define agentic actions with safe scopes: start with read → suggest → human-confirmed write, then expand as confidence grows.
What customers and analysts are seeing
Organizations report spending less time on debugging, coordination, and troubleshooting when using the developer experience capabilities like Live Debugger. That time shifts to building features with more confidence and fewer rollbacks.
Industry analysts are also pointing to frontend and mobile quality as a rising priority in AI-assisted engineering. Real User Monitoring, Error Inspector, and strong ANR/crash symbolization give practitioners live context where user experience is most fragile.
Availability
Some enhancements are available now; others will roll out over time. Dynatrace is running hands-on workshops, a vibe coding event, and a hackathon in partnership with Microsoft, powered by GitHub Copilot. Expect continued updates as the DevCycle integration progresses.
Related tech
- Claude by Anthropic for LLM-driven features and agent workflows.
- Azure AI Foundry for multi-model orchestration and deployment.
Upskilling your team
If you're formalizing skills around agentic systems, observability, and release engineering, here's a curated list of AI courses by job role: AI courses by job. Use it to align training with your product roadmap and stack.
Bottom line: Product teams don't just need more data-they need the ability to change system behavior safely, in real time. Dynatrace's updates push observability into that space, where control meets context and releases move with less drama.
Your membership also unlocks:
