With the addition of Now Assist to the Vancouver release of its software platform, ServiceNow is embedding gen AI across the three major workflows it supports. Now Assist for IT Service Management, Customer Service Management, and HR Service Delivery add new text creation and summarization features and an interactive chatbot interface to help workers get to relevant information more quickly.
ServiceNow has developed its own domain-specific large language model, Now LLM, to assist with these enterprise workflows, although enterprises can also hook up the new assistants with other commercially available models, or even their own.
The Vancouver release of Now Platform also includes new automations and security tools.
A return to the command line?
Modern enterprise applications have moved away from legacy green-screens to browser-based interfaces that can be accessed from PCs, tablets, or phones. The text-based chatbot interface used in so many gen AI implementations, including ServiceNow’s, is in some respects a step back for usability, a return to the command line that so many hunt-and-peck typists were glad to leave behind.
“So we’re taking a dual-track approach,” said Amy Lokey, ServiceNow’s SVP of product experience. “There are in-app contextual abilities to access generative AI, and there’s the analysis panel on the right, and our users may choose one or the other.”
The contextual abilities enable users to click on a button in Now Platform’s web interface to perform actions such as summarizing a case to date in Now Assist for HRSD or an agent’s chat with a customer to update a case record in Now Assist for ITSM.
A new analysis panel, present on the right of the screen in these workflows, is where users can interact with the chat interface. “As you navigate through applications on the platform, that layer of help stays with you,” Lokey said. “It can even help you navigate from one part of the platform to the other. Maybe typing isn’t the fastest thing to do, but it’s a pretty universal way to engage.”
One advantage of the text-and-typing interface, she said, is that it can be used with voice-to-text entry and screen readers.
There’s also support for enterprises that want to roll their own workflows. Now Assist for Creator is intended to help development teams build processes faster with its text-to-code capabilities. These can take in natural-language descriptions of functions and suggest code fragments, or even complete code.
“For all those things, when you have a domain-specific LLM, it provides a better result,” he said. “They’re actually faster, cheaper, and safer. When it comes to domain specific knowledge that we need for ServiceNow, we build our own LLM.”
Cost in question
The new gen AI features do come at a cost. Enterprises will have to pay for a “Professional Plus” or “Enterprise Plus” add-on pack for a given workflow, which will include a bundle of “assists,” or calls to the gen AI functionality, according to Sigler.
Assists will be counted regardless of whether an enterprise uses ServiceNow’s LLM, one hosted by a provider such as OpenAI, or one of its own, and once the bundle is used up, enterprises will have to pay again.
“We’re very generous when it comes to the number of assists that you get,” Sigler said. “Eighty to 90% of our customers are going to be just fine with the number of interactions that we allow with those LLMs.”
Sigler expressed confidence that the remaining 10% to 20% of ServiceNow’s customers facing additional usage-based fees would get their money’s worth, though: “The value to the customer is enormous here.”
For Stephen Elliot, IDC’s group VP for cloud operations and devops, cost isn’t a major stumbling block, as there are many other aspects of gen AI to consider first.
“CIOs need to have a specific business and governance plan to balance and accept AI’s risks and benefits,” Elliot said. “Product pricing is only one part of the discussion. Other pieces include staffing, skills training, partner involvement, legal exposure, and risk mitigation.”
The AI model used also has a big influence, he said: “Generic models can add some value, while domain specific models can add massive value.”
The Vancouver release of Now Platform and its Now Assist add-ons is scheduled for release on September 29, 2023.
Sigler said he thought ServiceNow might be the first enterprise software company to make gen AI generally available across its software platform. “We hear a lot of companies out there in alpha and beta, or whatever they’re going to call it,” he said. “We’re actually going to be making this available, generally available, and for sale on the 29th.”
He may have a point: while platform makers such as OpenAI have offered gen AI components for some time now, other enterprise software vendors have taken a more cautious approach. Last week Salesforce made a big announcement about its plans to make its Einstein Copilot gen AI assistant available across its software platform—but only to a limited number of customers, sometime before year-end.
It’s too early to say whether there’s any advantage to being the first mover, though, according to IDC’s Elliot. “This is going to be a marathon, not a sprint,” he said.
Vancouver’s other new features
Gen AI isn’t the only new feature in the Vancouver release, however. There are also new automations for health care providers and HR and finance teams, and a new approach to application security.
Zero Trust Access is a new feature for ServiceNow Vault, enabling enterprises to add granular authentication policies for access to their data, and evaluate risk based on location, network, device, and user, and limit access to data accordingly. ServiceNow has also expanded its third-party risk management tool, allowing more employees to use it to evaluate the security of applications.
For health care companies, a new automation process for clinical device management will help teams responsible for the maintenance of medical equipment to track their work and order necessary parts.
And for finance staff, ServiceNow has enhanced its Source to Pay Operations processes with the addition of Accounts Payable Operations. This will assist with digitizing the process of receiving invoices, their reconciliation, and payment. ServiceNow expects this will help reduce the cost of processing invoices from $16 manually to under $3, Lokey said.
Application Management, Artificial Intelligence, CIO, Enterprise Applications, Generative AI, ServiceNow, Software Deployment, Software Development, Software Providers, Zero Trust