Datadog has launched greater than 100 new merchandise and options centered on AI operations, safety and software program improvement. The rollout was introduced on the firm’s DASH occasion in New York.
The updates give attention to autonomous monitoring, safety for AI brokers, log administration in buyer cloud storage, customized agent creation and centralised oversight of AI agent use. The additions are meant to assist prospects handle operational complexity as AI instruments unfold throughout software program improvement and infrastructure.
A key change is the growth of Bits AI, Datadog’s set of AI brokers for improvement, safety and operations. The system now goes past root-cause investigation and might detect, examine and remediate points routinely throughout infrastructure and the software program manufacturing cycle.
Bits AI can constantly scan infrastructure, floor issues, advocate fixes and, in some circumstances, resolve them inside pre-defined controls. It may well additionally monitor releases and pull requests from code modifications and staging by means of to manufacturing, checking whether or not software program behaves as anticipated at every stage.
One other addition is Agent Evals, a function that may debug AI brokers and generate fixes for them. Bits AI can also be obtainable by means of instruments together with Slack and Claude.
Olivier Pomel, Co-Founder and Chief Government Officer at Datadog, set out the corporate’s view of the market shift. “AI has created new operational challenges the place code improvement has outpaced human-scale administration and malicious actors now use AI to assault crucial methods. However AI did not create this complexity – it accelerated what was already there. The businesses that win on AI will not simply construct higher fashions, they will construct operational management round them,” mentioned Pomel.
Datadog additionally launched AI Guard, a safety product geared toward AI brokers. The device is designed to detect and block assaults hidden in prompts or in multi-step agent behaviour, an space the corporate argues might be missed by conventional prompt-and-response testing.
Tim Knudsen, Vice President of Safety Merchandise at Datadog, described the dangers related to brokers given broader entry to methods and information. “With AI brokers working with elevated privileges, accessing delicate information, and speaking externally, a single malicious immediate hidden in an innocuous-appearing immediate can flip a well-intended agent right into a malicious actor leaking delicate info – costing tens of millions in reputational injury and information loss. However attackers have realized to cover agent poisoning utilizing refined directions solely detectable with a deep understanding throughout a number of steps of the agent’s behaviour. Datadog’s new AI Guard makes use of a novel mixture of deep agent telemetry tracing and AI-native stateful behavioural anomaly evaluation to detect and block AI agent assaults in any other case missed by stateless prompt-and-response analysis,” mentioned Knudsen.
Information storage
One other a part of the launch addresses the price of storing and looking out massive volumes of logs. Datadog’s Convey Your Personal Cloud choice permits prospects to deploy the platform in their very own surroundings so information might be processed and listed in cloud object storage they already management.
The corporate positioned this as a response to the sharp rise in log information linked to AI workloads. For a lot of organisations, the selection has been between retaining extra information at increased price or deleting a few of it and dropping visibility into methods.
Agent oversight
Datadog additionally launched Bits Agent Builder, which lets prospects create their very own AI brokers contained in the platform for remediation and operational workflows. These brokers might be configured to resolve incidents, generate studies and implement requirements inside customer-defined controls.
Alongside that, the corporate launched Agent Console, a monitoring product for AI brokers and developer instruments resembling Claude Code, Cursor and GitHub Copilot. The purpose is to offer groups a central view of adoption, efficiency and spending linked to agent use.
Alexis Lê-Quôc, Co-Founder and Chief Expertise Officer at Datadog, mentioned the launch was organised round visibility and management. “We constantly make investments about 30 per cent of income into R&D, which is why we’re in a position to deeply perceive and resolve the issues our prospects face day by day in managing operational complexity,” mentioned Lê-Quôc.
He added: “At DASH, we launched 100+ capabilities unified round one aim: giving prospects the visibility they should discover and repair the problems that matter most, the second they matter.”
Lê-Quôc additionally outlined how the corporate sees the subsequent stage of Bits AI. “To this point, Datadog’s Bits AI has centered on investigating the basis explanation for points. Now – with Bits Detection, Agent Evals, Infrastructure, Code, Launch, Information Evaluation, Testing and Chat – Bits AI is able to actually autonomous operations, changing into a dependable teammate that operates throughout each stage of the manufacturing lifecycle and improvement loop,” mentioned Lê-Quôc.
On the brand new monitoring product for coding brokers, he mentioned: “Whereas there is no such thing as a doubt coding brokers are rushing software program improvement, a scarcity of visibility makes it tough to know the complete influence these brokers have on the enterprise. Agent Console gives the wanted visibility to reply the important thing questions for customers concerning the heaviest adopters of brokers, the duties that brokers carry out finest and the place they wrestle, and the way the work produced by brokers correlates with spend,” mentioned Lê-Quôc.