Today on Product Saturday: Databricks introduces an open-source proposal for sharing corporate data, Datadog adds more bits to Bits AI, and the quote of the week.
Today: Anthropic's first Mythos-class AI model launch did not sit well with a lot of enterprise users, Oracle investors remain wary about its expansion plans, and the latest enterprise moves.
Today: Microsoft scrambles to minimize the fallout from a second batch of compromised open-source patches in a month, Apple teams up with Google for an expansion of its cloud security service, and the latest funding rounds in enterprise tech.
Today on Product Saturday: Databricks introduces an open-source proposal for sharing corporate data, Datadog adds more bits to Bits AI, and the quote of the week.
Welcome to Runtime! Today on Product Saturday: Databricks introduces an open-source proposal for sharing corporate data, Datadog adds more bits to Bits AI, and the quote of the week.
Please forward this email to a friend or colleague! If it was forwarded to you, sign up here to get Runtime each week.
Share the wisdom: Open data formats are one of the key building blocks of enterprise AI, and they have made it much easier for companies to feed their AI apps and agents with the right data and the right time. Ahead of its Data + AI Summit next week in San Francisco, Databricks introduced a new open protocol for sharing that data between companies who are partners, customers, or suppliers.
OpenSharing is an evolution of the Delta Sharing protocol that will be "the first open protocol to cover agent skills, AI models, and unstructured data," Databricks said in a press release. It "can publish agent skills and AI models via a single open protocol that any partner can consume, with standard APIs for discovery, authorization, and access, regardless of platform," and it will be hosted by The Linux Foundation.
Ducks on the pond: The rise of agentic coding tools is having a profound impact on the way software is built and maintained, and the same trends are also changing the way data engineers go about their business. This week MotherDuck, the company behind the DuckDB open-source analytical database, unveiled a new way for customers to connect their agents with their data.
Flights allows customers to "build and deploy data pipelines in minutes using a flexible, general-purpose Python runtime," the company said in a blog post. "AI has come for the data world. There’s a big opportunity to take things that were traditionally manual tasks and use prompting and the English language instead of having to hand-code everything," MotherDuck CEO Jordan Tignani told SiliconAngle.
Bits and bobs: Datadog is on a bit of a run right now, enjoying a 71% jump in its stock price for the year to date as investors realize observability tools could become even more important to enterprise customers as agents start running in earnest. This week at its DASH conference the company showcased eight new features for its Bits AI agent suite that will allow those customers to automate more of the work of using those tools.
"With these critical updates, Bits AI automatically detects, investigates and remediates issues by scanning infrastructure around the clock to surface issues, recommend fixes and resolve them," the company said in a press release. It also introduced AI Guard, which uses all the same telemetry data to detect prompt-injection attacks.
Issue skills: Linear's issue-tracking and product-management tools are the darling of the startup crowd, and after some early hesitation around AI the company has jumped into the deep end in recent years. This week it introduced a big update to its Linear Agent product that allows customers to point the agent at issues that need resolution.
LInear Agent can now write code; after a real live human being flags an issue, "the agent reads the issue and its surrounding discussion, investigates the codebase, proposes an approach, writes the code, and opens a pull request," the company said in a blog post. It also now supports diffs, which makes it much easier for people to tell where the agent has changed existing code when reviewing its output.
North star: Anthropic and OpenAI continue to dominate the AI model conversation as they hurtle toward enormous IPOs, but as enterprise concerns about costs and data start to mount, they'll start thinking about using other models at the heart of their apps and agents. Cohere has raised a modest $1.7 billion (by today's standards) to work on its own models for enterprise use, and this week it launched its first model designed for software developers.
"North Mini Code represents a step forward in small agentic coding models that can accomplish tasks that matter to developers," the company said in a blog post. It's available under the Apache 2.0 license, and a bet that more companies will soon be interested in a model "designed for speed and efficiency, with a strong focus on minimizing total cost of ownership as we continue to refine and scale the model," it said.
GitLab Orbit gives your AI agents the context to move at full speed. Smarter agents, faster results.
The CTOs and CIOs at big enterprise companies love to talk about how heavily they're investing in enterprise AI, but the CFO keeps score. According to new research from DoIT, just 15% of enterprise finance leaders think they understand how to calculate the ROI they're seeing from AI investments, while 74% of that group thinks their company has "mature" or "leading edge" AI spending controls.
"If you rely on public registries and repos, and your devs are using AI coding tools, you have a high likelihood of being impacted by this specific attack vector." — Cloudsmith's Nigel Douglas, issuing a warning to all enterprises that if Microsoft can be compromised twice in a short period of time by a new wave of supply chain software attacks targeting packages, so can your company.
Anthropic announced late Friday that the U.S. government ordered the company to suspend access to Claude Mythos 5 and Claude Fable 5 by "any foreign national," which appears to include Anthropic employees, forcing it to "disable Fable 5 and Mythos 5 for all our customers to ensure compliance," it said in a blog post.
GitHub is "now serving 40% of monolith traffic from Azure (up from 8% in February)," it said Thursday, acknowledging that the service is still having significant problems remaining online as it moves out of its own data centers.
OpenAI announced plans to acquire Ona, which operates a cloud platform service for running AI agents that seems like it will get folded into Codex, for an undisclosed amount.
A zero-day flaw in Oracle Peoplesoft went unaddressed by the company for two weeks despite being actively exploited until hours before Google's Mandiant published an advisory, according to The Stack.
GitLab Orbit gives your AI agents the context to move at full speed. Smarter agents, faster results.
Thanks for reading — next week is a big week for Runtime, stay tuned!
Today: Anthropic's first Mythos-class AI model launch did not sit well with a lot of enterprise users, Oracle investors remain wary about its expansion plans, and the latest enterprise moves.
Today: Microsoft scrambles to minimize the fallout from a second batch of compromised open-source patches in a month, Apple teams up with Google for an expansion of its cloud security service, and the latest funding rounds in enterprise tech.
Today on Product Saturday: Microsoft unveils a new open-source project designed to get apps running on Microsoft Fabric, Workday launches a new agent-building tools and observability service, and the quote of the week.
Today: Microsoft's Omar Shahine thinks knowledge workers will soon delegate a lot of their busywork to its new personal assistant, Anthropic details how Claude is building Claude, and the latest enterprise moves.
