“We bring our domain and knowledge as well as the context to build applications using Claude’s models trained with industry-specific content to build differentiated services,” said Amit Zavery, President, COO, and Chief Product Officer at ServiceNow.
ServiceNow on Wednesday announced a collaboration with Anthropic to strengthen the relationship between Anthropic Claude AI models and ServiceNow workflows.
Under the agreement, Claude will become the default model behind ServiceNow Build Agent, the company’s vibe coding technology that uses language inference models and search augmented generation (RAG) to translate plain language and business needs into code.
Amit Zaveri, president, COO, and chief product officer at ServiceNow, told CRN that the company has built an agent AI platform, partnered with multiple vendors, and chose cloud systems and large-scale language models to provide customers with a lot of automation.
[Related: ServiceNow Beefs Up Channel Program With AI Emphasis]
“We’ve also seen some unique use cases emerge where we have the potential to do even more with certain partners,” Zavery said. “As we expand our relationship with Anthropic, there is a feature we released last year called Build Agent, which is a vibe coding interface. We have always used no-code and low-code to build and extend the ServiceNow platform. Build Agent provides a vibe coding interface for building workflows or prompts.”
That code can run on the ServiceNow platform, which has enterprise compliance, security controls, guaranteed runtime performance, and more, all in place, Zavery said.
“To that end, we are increasingly using Claude as our underlying code generation tool and build agent,” he said. “And we’ve added a lot of features around things like enterprise security to ensure that it not only works very fast, but also has full enterprise compliance capabilities.”
Zaveri said customers are free to use any platform other than Claude for vibe coding.
“One of the things we saw with Claude is that it has very good code generation capabilities and the ability to build and generate code based on prompts,” he said. “We’ve been using it within ServiceNow for some of our engineering work. It’s more of an atmospheric coding environment than low-level coding, but we know that this generation is very good and powerful, and we can make it even more powerful using Build Agent.”
When asked about the issues enterprises face with Vibe coding, such as security risks, code quality, and scalability, Zavery said Claude with Build Agent alleviates those issues.
“When we build an application, we provide a data model,” he said. “When we build workflows, we generate a lot of data models based on the historical way we build workflows. So the foundation and scaffolding of those models actually prevents a lot of these errors and illusions. This is true when you use Build Agent and Claude on its own, when you use Claude on its own, or when you use Enterprise I think that’s a big difference when using other code generation tools out there for your use case. What we’ve done with this platform is track what the application looks like. How is it created? What is the data model? We actually guarantee that outcome based on our understanding of the field and our years of expertise.”
ServiceNow is seeing a number of interesting AI use cases emerging, particularly in regulated industries such as life sciences and healthcare, Zaveri said.
“Anthropic with Claude has done a lot of training on datasets, understanding specific industry domains and contexts,” he said. “We’re bringing these together to co-innovate around things like what we want to do in problem solving, investigative analysis, case management, claims, etc. We’re bringing our domain and knowledge as well as the context to build applications using Claude’s models trained with industry-specific content to build differentiated products.”
Additionally, Zavery said ServiceNow last year launched ServiceNow AI Control Tower, which manages the lifecycle of AI deployments, AI agents, and AI systems.
“We discover it, we manage it, we observe it, we track it,” he said. “We help from an outcome perspective, not just the lifecycle, but also security and the data associated with it,” he said. “We also now have the ability to discover Claudes, easily track them, and make them part of our ecosystem. So our customers can manage all their AI implementations and lifecycles through one product called AI Control Tower. It’s built on CMDB (Configuration Management Database) and scales very well. It does all the tracking for hardware and software. We use AI for tracking and management. By adding , our product becomes even more powerful and allows us to make sure that all the products and things that our customers might use are supported through AI Control Tower.”
