Today on Product Saturday: AWS introduces a new service for running AI agents developed with OpenAI, ServiceNow builds an agent for running agents, and the quote of the week.
Today: Cloudflare's Next.js rewrite project could be an early sign of a new era of custom software, a roundup of enterprise tech earnings, and the latest enterprise moves.
Today: Why Monday's software-stock selloff and Tuesday's software-stock rally shows how investors don't understand how and why enterprises use software, OpenAI's Project Stargate will be lucky to settle for the moon, and the latest funding rounds in enterprise tech.
Today on Product Saturday: AWS introduces a new service for running AI agents developed with OpenAI, ServiceNow builds an agent for running agents, and the quote of the week.
Welcome to Runtime! Today on Product Saturday: AWS introduces a new service for running AI agents developed with OpenAI, ServiceNow builds an agent for running agents, 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, and if you value independent enterprise tech journalism, click the button below and become a Runtime supporter today.
State of play: AWS has been chasing OpenAI pretty much from the day ChatGPT launched in 2022, and on Friday it chalked up its biggest win to date. Alongside plans for Amazon to invest up to $50 billion in OpenAI, the two companies announced a joint project to build a "stateful" development environment on Amazon Bedrock that would allow developers to build AI agents that have access to background details about incoming prompts.
The Stateful Runtime (!) Environment would eliminate the orchestration layer that "stateless" APIs need to understand how to answer questions so that "instead of manually stitching together disconnected requests to make things work, your agents automatically execute complex steps with 'working context' that carries forward memory/history, tool and workflow state, environment use, and identity/permission boundaries," OpenAI said in a blog post. However, as Axios pointed out, "that distinction matters because Microsoft has the exclusive right to deliver traditional API calls, which is where developers pay to query OpenAI services without sharing who is asking or for what purpose."
Storage locker: 2026 will be an interesting year for the neocloud phenomenon, as companies like CoreWeave and Lambda try to keep up with the massive investments that the hyperscalers are making in capacity. BackBlaze launched a new storage product this week called B2 Neo that it claimed was designed with those upstarts in mind.
B2 Neo allows neoclouds to offer storage services to their customers and "provision accounts, manage permissions, and handle billing through existing platform tools without a separate console or manual setup," BackBlaze said in a press release. Most neoclouds (how long does one get to be called "neo"?) understand they need to offer more than just GPUs-by-the-hour if they want to gain a real foothold in the enterprise, and are investing in software and other services to close the gap.
Ants marching: ServiceNow has been one of the poster companies for the SaaSpocalypse over the last few months, whether or not it actually earned that position. This week it introduced a new service called Autonomous Workforce that aims to show how AI agents can improve its existing products, rather than rendering them obsolete.
Autonomous Workforce "orchestrates teams of AI specialists with roles such as a Level 1 Service Desk AI Specialist, Employee Service Agent, or Security Operations Analyst to execute work from start to finish," the company said in a press release. The company said the service is already "handling 90%+ of employee IT requests" inside ServiceNow, and claimed the Service Desk AI Specialist can handle inquiries "99% faster than when these cases are handled by human agents."
Matter of trust: Now that agents are finally starting to get a foothold in the enterprise, ServiceNow isn't the only company working on ways to coordinate their activity. This week VAST Data introduced two new services that promise to help customers manage agent activity and tune AI models for better results.
PolicyEngine "governs agents’ access to shared memory, external tools, knowledge bases, or other agents," while TuningEngine "captures outcomes from agentic pipelines and utilizes curated feedback to enhance model performance over time," the company said in a press release. "Our conclusion was if we don’t handle fine-tuning, then that’s going to be a security gap that ultimately makes AI less trustable,” said VAST co-founder Jeff Denworth, according to The New Stack.
Agent tracking: Software-development organizations have been in a love-hate relationship with Atlassian's Jira for decades, but it remains one of the most widely used products in enterprise tech. This week the company announced that the project-management tool now supports agents from third-party software vendors as well as homegrown versions.
Jira users can now tag agents in their normal workflows to assign projects and keep track of their progress. "Because agents operate inside Jira’s existing structures, they respect your permissions, project configurations, workflows, and audit trails," the company said in a blog post.
Identity-management tools are going to be even more important if agentic AI really takes off inside the enterprise, according to new research from Sophos. Two-thirds of all incidents that Sophos worked on last year involved "identity-related attacks," although on the positive front, the mean time to detect and respond to those attacks declined to three days.
"The reality is that these new AI tools will not rip and replace existing software ecosystems and data environments with these AI tools only as useful as the data it can reach." — Wedbush Securities analyst Dan Ives, attempting to inject a dose of reason into the stock market's current view on enterprise software, which he correctly describes as dumb.
Secretary of Defense Pete Hegseth declared Friday that Anthropic is a "supply-chain risk to national security" after the company refused to let the Pentagon violate the terms of service of its AI models. He added that "no contractor, supplier, or partner that does business with the United States military may conduct any commercial activity with Anthropic," which would appear to disqualify pretty much every single U.S. software company from military contracts.
But Anthropic said in a blog post that "the Secretary does not have the statutory authority to back up this statement," and that it would challenge the whole thing in court, assuming Hegseth actually puts down the Twitter machine and designates the company as a supply-chain risk through the proper channels.
"Designating Anthropic as a supply chain risk would be an unprecedented action — one historically reserved for US adversaries, never before publicly applied to an American company," it said in the statement. "No amount of intimidation or punishment from the Department of [Defense] will change our position on mass domestic surveillance or fully autonomous weapons."
Thanks for reading — see you Tuesday!
Today: Cloudflare's Next.js rewrite project could be an early sign of a new era of custom software, a roundup of enterprise tech earnings, and the latest enterprise moves.
Today: Why Monday's software-stock selloff and Tuesday's software-stock rally shows how investors don't understand how and why enterprises use software, OpenAI's Project Stargate will be lucky to settle for the moon, and the latest funding rounds in enterprise tech.
Today on Product Saturday: Anthropic freaks out the enterprise security market with a new Claude feature, Tailscale extends its networking security tech to agents, and the quote of the week.
Today: Members of the Runtime Roundtable share their tips and tricks on getting agents from experiment to production, Google drops a new high-end version of Gemini, and the latest enterprise moves.
