Salesforce hits back at ServiceNow; Zuora helps you get paid

Welcome to Runtime! Today on Product Saturday: six months after ServiceNow launched a product taking on Salesforce's core market, the "ohana" strikes back, Zuora sees big changes coming to SaaS pricing, and the quote of the week.

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

Ship it

ITSMs, on guard: The highlight of ServiceNow's Knowledge 2025 event earlier this year was a new CRM service aimed squarely at rival SaaS giant Salesforce, and this week Salesforce decided to return the favor. Agentforce IT Service is "an agent-first, conversational-first IT support product suite," the company said in a press release that starts off complaining about the "frustrating productivity drains brought on by ticket-based IT Service Management (ITSM) processes."

Of course, just like every other enterprise software company, ServiceNow is scrambling to put agents at the core of its flagship ITSM product, but it's a sign that old competitive boundaries are falling away thanks to the AI boom. "We’ll absolutely run into ServiceNow and we are very confident of our innovation — and our continued innovation — to be able to take that," Salesforce's Kishan Chetan told Bloomberg.

Our menu options have changed: Outside of coding, generative AI and agents are probably having their biggest impact on the customer service department, in hopes that the new technology can solve customer issues on its own before that customer starts screaming "OPERATOR" into the phone. Zendesk has sold a lot of software into those organizations over the years, and this week it introduced new agents for its contact-center product that it promised can solve problems without forcing customers through a terrible phone tree.

The new lineup includes voice agents that were "designed to understand natural speech, take action, and resolve issues without escalation," as well as live video support, which "enables agents to switch to high-touch interactions, seeing exactly what customers see to resolve complex issues with empathy and build trust in new markets," the company said in a press release. Zendesk's Shashi Upadhyay told TechCrunch that the new agents can resolve 80% of the issues they encounter without escalating to a human, and that sounds like a bet.

Every workload has its price: Generative AI bots and agents have raised a lot of questions about pricing strategies for SaaS companies, because the tried-and-true subscription or per-seat models don't make as much sense when so much of the work is supposedly being automated. Zuora's suite of financial-services software is evolving along with everything else, and this week it introduced new products that it said can help companies transition to consumption-based pricing models.

The new services include better analytical tools for helping SaaS companies understand exactly how customers are using their products, and a feature that allows those companies to help their own customers understand their usage patterns. "As customers learn how to interact with and pay for AI-driven products, monetizing value in real time and usage visibility become even more important, but create greater complexity for finance teams," the company said in a press release.

Network = computer: After Dell's bullish forecast on the future of investment in on-premises data-centers this week, Cisco rolled out a new router designed to handle AI training workloads. The Cisco 8223 is "the industry’s first 51.2-terabit fixed Ethernet router, featuring the highest bandwidth 51.2-terabit deep-buffer routing silicon," the company said in a press release.

The new routers were designed to help coordinate AI training runs across data centers spread out over large distances. "For a bunch of different reasons, this scale across network is absolutely vital to the scalability of AI workloads," according to The Next Platform, noting that Cisco will be butting up against Nvidia and Broadcom as it goes after this market.

Self-cleaning coding machines: Security teams have been excited about the potential of generative AI technology to help automate the tedious task of finding security problems in software before it ships, and Google DeepMind unveiled new research this week that might get them a little closer to that goal. CodeMender is an AI agent that takes "a comprehensive approach to code security that’s both reactive, instantly patching new vulnerabilities, and proactive, rewriting and securing existing code and eliminating entire classes of vulnerabilities in the process," DeepMind said in a blog post.

DeepMind emphasized that CodeMender is still a research project, and no patches generated by the agent are submitted to any code base without human review. However, "we've already begun submitting patches to various critical open-source libraries, many of which have already been accepted and upstreamed," DeepMind said, and if this research turns into an actual reliable service it could go a long way toward solving the open-source software supply-chain security problem.

Stat of the week

APIs are the lifeblood of enterprise software; they allow developers to build rich connections between different applications, and it's hard to understand how this world would work without them. And like everything else in this world, AI agents are taking over: "Nearly one in four developers (24.3%) are already designing APIs with AI agents in mind, a fundamental shift that signals the rise of machine-consumable APIs," according to a new report from Postman.

Quote of the week

"You should expect a huge focus from us on really leaning into enterprise." — OpenAI CEO Sam Altman, signaling his company's intentions to get much more serious about enterprise AI and further complicating its relationship with Microsoft.

The Runtime roundup

Stocks fell sharply Friday afternoon after President Trump declared new tariffs of 100% on basically anything coming out of China and also vowed to impose export controls on "any and all critical software," which ruined the weekends of enterprise software executives across the country.

The chip nerds (complimentary) over at Semi Analysis introduced a new benchmark for GPU inference that will be "continually re-benchmarking the world’s most popular open-source inference frameworks and models to track real performance in real-time," and it will hopefully bring some transparency to the otherwise murky world of chip benchmark gamesmanship.

Thanks for reading — see you Tuesday!