Dynatrace Intelligence Explained: What Replaced Davis AI
If you knew Dynatrace's AI as Davis, the name has changed and so has the scope. Dynatrace Intelligence is now the umbrella for all the AI in the platform, from the causal engine that finds root cause to agents that investigate incidents for you. Here is what it is, what each part does, what you need to use it, and how it is charged.
Our position, before you read on. C4C Group is an official Dynatrace partner, and Dynatrace is the observability platform we usually recommend for large organisations. This guide is based on Dynatrace's own documentation and announcements, and we say where a feature is still in preview.
Dynatrace Intelligence is the name for all the artificial intelligence in Dynatrace, and it replaces the Davis AI name. It combines deterministic, causal AI, which works out root cause from a live map of how your systems depend on each other, with predictive AI, generative AI (Dynatrace Assist, formerly Davis CoPilot) and agentic AI that can investigate and act. Dynatrace launched it at its Perform conference in January 2026 and describes it as an agentic operations system that fuses deterministic and agentic AI. The generative and agentic parts need Dynatrace SaaS on the latest platform and a DPS subscription.
What changed from Davis AI
Davis was the name Dynatrace used for its AI engine for years, best known for automatic root cause analysis. Dynatrace now uses Dynatrace Intelligence as the umbrella term for all AI in the platform, including generative, agentic, causal and predictive AI, and says the new term replaces the old name and reflects expanded capabilities. Dynatrace, AI presence. Davis CoPilot, the generative assistant, became Dynatrace Assist. You will still see Davis in some documentation and in terms such as Davis events while the naming settles.
The four kinds of AI inside it
| Kind | What it does | Where you see it |
|---|---|---|
| Causal (deterministic) | Follows the dependency map to the component that started a problem, and groups related alerts into one problem | Problems, root cause analysis |
| Predictive | Learns baselines, including seasonal patterns, and forecasts trends such as capacity | Anomaly detection, forecasting |
| Generative | Turns plain English questions into DQL queries and explains findings | Dynatrace Assist |
| Agentic | Agents that investigate incidents, annotate problems with findings and recommendations, and can be built for your own workflows | SRE, developer and security agents, Agent Builder |
Dynatrace lists the underlying models in its documentation, including causal correlation analysis, predictive analysis, a seasonal baseline model and a DQL generation model. Dynatrace, AI models.
Why the causal part matters
Most AI for operations works by correlation: alerts that fire together are probably related. Dynatrace's approach is different because the platform already knows the topology. OneAgent® discovers hosts, processes and services, Smartscape® keeps a live map of their dependencies, and Grail® holds the telemetry in one place. With that map, the AI can trace a slow checkout back to the database or deployment that caused it, rather than presenting a cluster of red alerts. Dynatrace says its AI grounds every action in deterministic, real time system understanding, in contrast with approaches that rely mainly on probabilistic outputs. Dynatrace, July 2026 announcement.
That matters even more once agents are involved. An agent that acts on a guess can make an incident worse. An agent that acts on a known dependency map and a deterministic root cause is far easier to trust with real work.
Ask what the AI knows before it answers. A language model on its own knows only what is in the prompt. Dynatrace Intelligence starts from the live dependency map and the full telemetry in Grail, and uses generative AI to explain and act on that, not to guess at it. That is the difference worth paying attention to when you compare AI for operations.
The agents, and what is available now
Dynatrace introduced domain specific agents for SRE, developer and security work at Perform in January 2026. In July 2026 it made the Cloud SRE Agent and an enhanced Dynatrace Assist available to SaaS customers on DPS, and announced an Autonomous SRE Agent, which picks up newly detected problems on its own, and Agent Builder, for creating custom agents without code. Dynatrace, autonomous operations announcement. Some agentic workflows are still labelled preview in the documentation, so check the status of any specific capability before planning around it.
The Dynatrace MCP server, which lets AI assistants and agents outside Dynatrace query it safely, is generally available. And with the acquisition of Arize, completed in October 2026, Dynatrace has said Arize's AI and LLM observability and evaluation capabilities will be integrated into the platform over time. Observability for AI workloads.
What you need to use it
- Causal and predictive AI are part of the core platform and have been for years.
- Generative and agentic AI are available to Dynatrace SaaS customers on the latest platform, and are not available on Dynatrace Managed. Dynatrace, agentic and generative AI FAQ.
- A DPS subscription is needed for licensed use. Dynatrace states that from version 1.347, accounts on older licensing models can no longer use the MCP server or generative and agentic workflow actions.
- Data privacy: Dynatrace states that customer data and prompts are not used for training.
How it is charged
Using Dynatrace Assist in the interface, or generative features without running queries, carries no licensing cost, although any query it generates and runs consumes query capacity under your rate card. Generative and agentic work run through workflows, the API or the MCP server is metered in AI units, a normalised measure of AI work weighted by task complexity, and drawn from your DPS commitment. Usage can be tracked in Account Management and on a ready made agentic AI usage dashboard. Dynatrace, AI units; Dynatrace licensing from C4C.
How we help
We are an official Dynatrace partner. We implement Dynatrace so the causal AI has what it needs, which means proper coverage, clean service naming and tags, and logs and traces joined up in Grail, and we help teams adopt the agents in a controlled way, starting with investigation before automation. If you are on classic licensing or Dynatrace Managed and want the newer AI, we will set out what the move to DPS and SaaS involves. Dynatrace implementation.
Want Dynatrace Intelligence working on your estate?
Tell us what you run today and whether you are new to Dynatrace, on classic licensing or already on DPS. We will tell you what it takes to put the AI to work and what to switch on first. We supply, implement and optimise Dynatrace.
Prefer email? Reach us directly at hello@c4cgroup.co.uk.
Frequently asked questions
What is Dynatrace Intelligence?
Dynatrace Intelligence is the umbrella name for all artificial intelligence in Dynatrace, covering causal, predictive, generative and agentic AI. It replaced the Davis AI name and was launched at Dynatrace Perform in January 2026, where Dynatrace described it as an agentic operations system that fuses deterministic and agentic AI.
Is Davis AI the same as Dynatrace Intelligence?
Dynatrace Intelligence replaces the Davis AI name and covers more. Davis was best known for causal root cause analysis, which is still at the core. Dynatrace Intelligence adds generative AI through Dynatrace Assist, formerly Davis CoPilot, and agentic AI through SRE, developer and security agents.
What is the difference between causal AI and generative AI in Dynatrace?
Causal AI is deterministic: it uses the live dependency map from Smartscape to trace a problem to the component that caused it. Generative AI, through Dynatrace Assist, turns plain English questions into DQL queries and explains findings. Dynatrace combines them so agents act on known root causes rather than probabilistic guesses.
Does Dynatrace Intelligence cost extra?
Causal and predictive AI are part of the core platform. Using Dynatrace Assist in the interface without running queries carries no licensing cost, but queries it runs consume query capacity. Generative and agentic work through workflows, the API or the MCP server is metered in AI units drawn from your DPS commitment.
Can we use Dynatrace Intelligence on Dynatrace Managed or classic licensing?
The generative and agentic capabilities are available to Dynatrace SaaS customers on the latest platform and are not available on Dynatrace Managed. Licensed use requires a Dynatrace Platform Subscription, and accounts on older licensing models can no longer use the MCP server or generative and agentic workflow actions.