Thoughts on Davis CoPilot capabilities compared to 'Agentic' workflows

[rgarzon1]

Looking at the market (Datadog Bits, AWS DevOps), the trend is clearly moving towards AI Agents that can perform actions, change configs, or manage incidents.

Currently, Davis CoPilot feels very limited to just DQL generation and data fetching. It feels rigid. To get any real action done, we still rely heavily on manual Workflow setups.

Is there a roadmap to bring real "agency" to Davis? It feels like we are lagging behind competitors who are turning their AI into active team members rather than just passive observers.

Or perhaps I’m missing something? Maybe there is a way to achieve this interactivity today that I haven't seen yet. If so, please let me know.

[shohebakthar]

Dynatrace is already moving toward real AI “agency” through Dynatrace Intelligence Agentic Workflows (Preview), which let Davis go beyond DQL and actually interpret signals, summarize issues, and trigger automated actions inside Workflows. With version 1.331+, you can chain Dynatrace Intelligence with existing workflow . Learn more here: https://docs.dynatrace.com/docs/shortlink/agentic-workflows and https://docs.dynatrace.com/docs/shortlink/preview-program#dynatrace-intelligence-agentic-workflows

[rgarzon1]

After some hands-on testing with the agentic copilot 🤔,

my takeaway is that it currently functions more as a conversational guide than a proactive engine. It lacks the autonomous reasoning to correlate metrics and logs without being prompted. I am aware of the new Agentic Workflows being introduced, which definitely show potential; however, they still appear to be focused on high-level reporting logic. We are still a few steps away from a truly independent 'investigation agent' that can rationalize gaps in data on its own.😀