There's an agent for that

Welcome to Runtime! Today: How AWS wants to be the center of agentic AI, just like everybody else, Microsoft hikes Office prices (again), and the latest enterprise moves.

(Please forward this email to a friend or colleague! And if it was forwarded to you, sign up here to get Runtime each week.)

Bias for agents

LAS VEGAS — While the vast majority of AWS's revenue still comes from its basic compute and storage infrastructure services, it has built a massive amount of software over the last two decades designed to help companies build, deploy, and manage their applications on its cloud. And as the entire world of enterprise software plunges headlong into the agentic AI era, tools that allow customers to build, deploy and manage AI agents could play an outsized role in making sure those customers stay on its bread-and-butter infrastructure services.

AWS introduced several new services this week at re:Invent 2025 in hopes of jump-starting agents as the future of enterprise software, joining basically every other company in this world betting on agents as the new SaaS. Those services come in several different flavors: tools that help companies build their own agents, and pre-built agents that customers can deploy against their AWS-hosted applications.

• AWS introduced Bedrock AgentCore in July to help companies build their own agents on the platform without having to provision infrastructure, and this week it unveiled two new services that promise to help customers deal with one of the most common complaints about agents: their tendency to produce unpredictable results.
• Policy allows companies to set additional guardrails governing what agents are allowed to do, and Evaluations helps them monitor the quality of their output before it reaches customers.
• Right now, "you can embed policies inside directly in your agent's code, but because agents generate and execute their own code on the fly, these safeguards are really best effort and can only provide you weak guarantees, and they're really difficult to audit in practice," said CEO Matt Garman during his Tuesday keynote.

But there are lots of companies that might be interested in using agents but don't want to deal with the hassle of creating and managing their own creations, just as later-arriving AWS cloud customers flocked to managed services. This week AWS unveiled several "frontier agents" designed to automate several key parts of almost every AWS customer's experience on the platform, as well as an agent that could help eliminate their tech debt.

• Customers will be able to assign software-development tasks to a new autonomous agent in AWS's Kiro coding assistant, use a new security agent to detect problems with their code before it hits production, and employ a DevOps agent that promises to help companies avoid costly outages.
• When it comes to DevOps, "agents help solve a lot of the traditional observability cloud operations, but they also generate voluminous additional telemetry now, because they're working 24/7," said Nandini Ramani, vice president of search and observability at AWS, in an interview with Runtime.
• And as Microsoft's Jay Parikh noted last month at Microsoft Ignite, coding agents have the potential to help companies power through the slog of modernizing old code and old systems in a fraction of the time it would take to accomplish that task manually.
• AWS Transform is the cloud leader's tool for tackling that problem, and this week it introduced new custom capabilities so "that you can teach it about your particular technology," such as internal software libraries written long ago that customers want to rebuild around a newer, more capable programming language, said Sriram Devanathan, director for agentic AI and software modernization, in an interview.

Right now every enterprise software company in the world is scrambling to sell customers on their approach to agentic AI for a clear reason: Once those customers build agents around a particular vendor's tools or embrace their homegrown services, it will be really hard to take their business elsewhere.

• AWS has a distinct advantage given that such a large number of companies already use its services, and after flailing about for a few years trying to figure its role in the generative AI era seems to have found its footing by realizing that customers will need a lot of help putting the pieces together.
• "These agents work in non deterministic ways, which is part of what makes them so powerful, but it also means that the foundation and tools that got us to where we are building software aren't necessarily the ones that we need for agents," Garman said Tuesday.
• If AWS is going to hold off Microsoft and Google — which have a lot of AI momentum right now — Garman needs to make sure AWS rebuilds that software foundation.

If you value independent enterprise tech journalism from big events like AWS re:Invent, please consider becoming a monthly supporter of Runtime. For $10 a month, you'll help us continue our mission to bring reliable and actionable coverage of this vital sector of the economy and gain access to supporter-only features currently in the works, such as an exclusive discussion and events forum.

Inflation, amirite

It's far from clear how many people actually want to use the AI features in Microsoft's Office and 365 suites, but it's very clear that providing those AI services is expensive. Microsoft announced Thursday that starting next July it will add new AI capabilities into the core of its flagship office productivity suite and as a result, prices are going up.

Almost every flavor of its Microsoft 365 Suite will cost around $2 or $3 more per user per month next year, the company said in a blog post. "Today we are announcing expanded availability of AI, security, and management capabilities coming to Microsoft 365 offerings in 2026," it said, framing the price increase as commensurate with those snazzy new features, but it's not like you get a choice whether or not you actually want to use them.

Microsoft last increased prices on the suite in 2022, according to Directions on Microsoft, and it still costs an additional $30 per user per month to tap into its most advanced AI features. Google also increased prices on its Google Workspace product earlier this year after force-feeding new AI features to customers, who have to work really hard to turn them off (ask me how I know).

Enterprise moves

Ken Yagen is the new chief product officer at CData, joining the data management company after product leadership roles at Warburg Pincus and Mulesoft.

Joyce Kim is the new chief marketing officer at Proofpoint, following marketing leadership roles at Zscaler, Twilio, and Genesys.

Ryan Shopp is the new chief marketing officer at Camunda, joining the process management company after marketing leadership roles at mabi and Deep Instinct.

Scott Torrey is the new chief revenue officer at Smartsheet, following sales and corporate leadership roles at Payscale and SAP Concur.

Kim Seabrook is the new chief revenue officer at Outsystems, joining the low-code development company after sales leadership roles at Shopify, Pendo.io, and Salesforce.

Jessica Ross is the new chief financial officer at GitLab, following finance and executive leadership roles at Frontdoor and Salesforce.

The Runtime roundup

Up to 39% of cloud users could be affected by a maximum-severity flaw disclosed Wednesday in React, but upgrading to newer versions of the JavaScript library should fix the problem.

Snowflake's stock fell more than 11% Thursday after it reported quarterly earnings that beat Wall Street's expectations but delivered a profit forecast that was weaker than expected.

Salesforce's stock went in the opposite direction after issuing revenue guidance for the current quarter that exceeded expectations.

HPE missed revenue estimates after an AI server deal was delayed and it also provided weaker-than-expected guidance for revenue in the current quarter.

Thanks for reading — see you Saturday!