2025 Keynote

Bits AI SRE acts as your always-on-call engineer that autonomously investigates alerts, forms and tests hypotheses using real-time telemetry, and recommends next steps before humans even log in. Fully integrated into the Datadog platform, Bits accelerates root cause analysis, reduces on-call fatigue, and helps your team resolve incidents faster by combining structured AI reasoning with complete observability context.

Make faster progress on alerts with the Voice Interface interface for Incident Response. Instead of a robotic phone call, get a real-time experience where you can get details of underlying monitors, get runbook next steps, and take actions while you’re still on the phone.

Jump right into digging into issues with Handoff Notifications for On-Call and Incidents. You can also dock incidents to see all the messages and graphs your teammates are sharing in your preferred chat app. Graphs can be compared across all pages while the dock is open

Keep users and teams informed during incidents with integrated status updates, from the same platform where you monitor, investigate, and resolve them. Built directly into the Datadog app, Status Pages let you communicate service health clearly and confidently by linking incidents to components, posting updates across lifecycle stages, and displaying historical uptime. Whether public or private, pages are fully customizable and easy to manage, so you can keep stakeholders in the loop without breaking focus.

Bits AI Security Analyst can now autonomously triage Datadog Cloud SIEM signals. Integrated directly into Datadog's platform, it conducts in-depth investigations of potential threats and delivers clear, actionable recommendations.

With the Action Interface, teams can trigger safe, audited actions on their infrastructure directly from chat using Bits AI—no workflows or apps required. Reliability (SRE), on-call, platform, and security engineers can restart services, reboot servers, flush caches, quarantine user accounts, and more in seconds in a safe and controlled way. Every action is validated against role-based policies and logged to Datadog Audit Trail, ensuring accountability.

The Bits Dev Agent is Datadog’s generative AI coding assistant. It merges knowledge of your observability data with source code context to diagnose issues, generate code, and create pull requests for multiple Datadog products, including Error Tracking, Code Security, APM, and more.

Pinpointing the origin of increased response times in distributed systems often requires manually stitching together data from traces, metrics, logs, and profiling to validate hypotheses about downstream bottlenecks, deployment changes, or database slowdowns. The APM investigator helps you diagnose and resolve application latency issues through a guided, step-by-step investigation workflow. It consolidates relevant analysis sections into a single interface, reducing the time needed to identify the root cause of the issue, without combing through raw telemetry data.

Proactive Recommendations automatically identify and surface the most impactful opportunities to improve application performance and reliability. Powered by real-time telemetry from APM, RUM, and Continuous Profiler data, these insights help pinpoint backend, frontend, and code-level inefficiencies. Recommendations are tailored to your team’s services and integrated directly into existing workflows so you can identify issues and improve your services’ performance and reliability faster.

Ship quickly and confidently with developer self-service, delivery guardrails, and live system data.

Build full-stack apps through natural-language interactions with Bits AI. Explain a need—for example, “Build me an app to revert deployments in GitLab”—and Bits AI auto-generates UI components, actions, and logic, all mapped to your existing data and permissions. Iterate by chat to tweak fields, flows, or styling without touching code, accelerating delivery while preserving the ability to dive into the underlying definition at any time.

The Remote Datadog MCP Server provides your Al agents access to Datadog context and features. You will be able to prompt Datadog in plain language right from Cursor, Codex, Claude Code, or your own Al agent "Why did latency spike on my service?" and get instant graphs, logs, and suggested fixes in-line with your code and in your work environment.

Eliminate the operational overhead of maintaining external storage, and context switching to search logs depending on their age. Flex Frozen lets you store your logs fully managed in Datadog for up to 7 years so you're always ready to comply with audits or investigate security breaches.

With Archive Search, run queries across logs in Flex Frozen or external archives — and access results directly, without rehydration.

Datadog Sheets offers a flexible and intuitive way to manipulate large volumes of telemetry data, including logs, RUM events, infrastructure metrics, and cloud cost data. You can create calculated columns, sort and filter rows, aggregate your data across multiple dimensions using a pivot table and visualize your results, all within a familiar grid view

Datadog Notebooks now supports advanced analysis features that were previously only available in Logs Workspaces. You can now flexibly query, join, and transform telemetry data such as logs, RUM events, spans, and reference tables directly within Notebooks. Use SQL, pre defined transformations and chain queries in order to dig into complex issues and find insights you can then easily share and visualize.

Bits Data Analyst Agent brings natural language and AI-driven guidance directly into Datadog Notebooks, enabling users to explore data, generate visualizations, and investigate issues conversationally. Instead of writing complex queries or stitching together SQL and transformations, users can ask rich analytical questions like: “Give me all the error logs from the past day and show me distinct log messages.

Migration to Datadog Log Management is made simple with automatic translation of your query to Notebooks. Bring Your Own Query from your existing logging tool and watch it be reconstructed in a step-by-step breakdown with full transparency.

Prevent sensitive data leaks in training datasets and prompt/response pairs with Sensitive Data Scanning. Support for more detections including sensitive data leaks in API metadata and other data poisonining attacks are coming later in 2025.

Ensure your AI model is safe and isn’t being manipulated against supply chain and model hijacking attacks, using Datadog Security. Support for more detections including model drift, model extraction, and jailbreaking are coming in the near future.

Learn more about our Workload Protection and Cloud SIEM products.

Protect your environment from code to cloud with Datadog Security. Support for more detections including agent tool misuse, novel identity attacks, and denial of service are coming later in 2025.

Learn more about our Code Security and Cloud Security products.

GPU Monitoring lets you track and optimize the utilization, cost, and performance of your GPU-enabled hosts, chips, and pods. This allows you to make informed provisioning decisions for GPU resources, and troubleshoot GPU-related issues affecting your high-compute applications quickly and accurately.Participation in this Product Preview requires an Infrastructure Monitoring license and ability to install the Datadog Agent on GPU Hosts. We will reach out within 2 weeks. Customers will be selectively admitted to the Preview based on broad alignment of their GPU monitoring needs and our offering roadmap.

With Datadog LLM Observability, teams building LLM applications can create ground truth datasets and run experiments to confidently validate changes to models, prompts, or code before deploying to production. For agentic applications, LLM Observability makes it easy to troubleshoot and debug the execution flow of agents—including decisions, tool selections, and hand-offs between agents.

With Datadog LLM Observability, teams building LLM applications can create ground truth datasets and run experiments to confidently validate changes to models, prompts, or code before deploying to production. For agentic applications, LLM Observability makes it easy to troubleshoot and debug the execution flow of agents—including decisions, tool selections, and hand-offs between agents.

AI Agents Console transforms your agentic ecosystem of custom-built agents and external, third party agents from a potential risk to your competitive advantage. With AI Agents Console, you can monitor the behavior and interactions of any AI agent that’s a part of your enterprise stack, whether that’s a Computer Use agent like OpenAI’s Operator, IDE agent like Cursor, DevOps agent like Github Copilot, Enterprise Business agent like Agentforce or your internally built agents. You'll have full visibility into every agent's actions, insights into the security & performance of your agents, analytics on user engagement and measurable business value from every agent.

With the rise of AI and the widespread adoption of platforms like Snowflake and Databricks, trust in data has never been more critical to business growth. Data Observability uses machine learning–based anomaly detection and end-to-end lineage to detect, resolve, and prevent data issues before they impact downstream BI and AI systems.