Solving enterprise AI’s last mile problem

Intelligent workflows translate AI predictions and recommendations to actions that drive business value

  • Companies struggle to convert machine intelligence into actions, at scale
  • AI and analytics that run natively in workflow platforms can solve this ‘last mile’ problem
  • AI-powered virtual agents use intelligent workflows to diagnose issues, offer recommendations, and take action

In popular culture, “artificial intelligence” (AI) is a broad concept that encompasses everything from computers that beat humans in games of chess and Go to virtual assistants that help us shop and play our favorite music. The term also covers algorithms that provide search results and recommend relevant content. It even includes Hollywood fantasies of malevolent cyborgs crushing humanity under their digital heels.

In the enterprise, AI generally refers to software systems that use component technologies like machine learning, intelligent automation, content understanding, and predictive analytics to deliver better, smarter experiences for employees and customers. These systems learn from data and improve over time.

Today, enterprises use disparate AI tools to understand content, conversations, machine logs, and so forth, and make recommendations on what action to take next in a given context. In order to deliver business value, these recommendations must be translated into actions, using workflows that deliver business value.

Yet many enterprises struggle to translate AI recommendations into actions (and hence, business value) in a consistent, scalable manner. At ServiceNow, where I serve as chief AI officer, we’re focused on solving this “last mile” problem of enterprise AI.

[Learn about our Element AI acquisition]

The expression comes from the transportation and telecom industries, where it refers to the challenge of moving people, goods, and data to their final destinations. In the enterprise AI context, the “last mile” is about taking action on AI predictions and recommendations, using workflows that drive business outcomes with measurable ROI.

ServiceNow is committed to empowering enterprises with AI-powered workflows that help employees work smarter and faster, streamline business decisions, and unlock new levels of productivity. With practical, purpose-built AI and analytics that run natively in the workflow platform, we enable employees to quickly make decisions, solve problems, find information, and automate tasks—making their work lives simple, easy, and convenient.

AI can predict and compare real-time patterns across operational systems, correlate data, and resolve issues before they impact customers or employees. Finally, AI can understand context and risk before an action is taken, predict resources needed, and advise on next steps.

[Read a statement from Element AI founder and CEO Jean-François Gagné]

Smarter virtual agents

Here are some practical examples. Let’s say I’ve forgotten my login credentials, or I’ve spilled coffee on my keyboard and need a replacement. Instead of calling the IT help desk and spending the afternoon on hold, I simply go to a service portal and engage with a ServiceNow Virtual Agent.

I type my problem in the chatbot window using plain English, or whatever language I happen to speak. I can use natural language phrases like “Forgot my password” or “My keyboard is damaged.” In all these cases, the natural language understanding capability in the Virtual Agent determines who I am, what I’m talking about (lost password or a damaged keyboard) and what I need (replace password or order a new keyboard).

The Virtual Agent then walks me through the steps needed to change my password or replace my gear. Finally, it triggers an automated workflow that sources the item I need and ships it to me at the address it has on file. (During this pandemic, that would be my home address.) It all happens using AI capabilities natively in the workflow platform itself.

In this example, AI understands the intent of my interaction (in plain English) with the Virtual Agent. It then reasons, placing its understanding in the proper context to infer what needs to be done next. Finally, it takes action to trigger a workflow that resolves my problem. This allows me to focus on strategic work that generates value for ServiceNow.

[Learn more about ServiceNow’s AI strategy and roadmap]

Human-machine teaming

Similar patterns apply to service delivery across all functions in an enterprise. Essentially, any repetitive task that involves some understanding of content, and reasoning in context to infer what needs to be done next, can be automated using AI-powered workflows.

Many business activities involve repetitive tasks, whether you’re running a customer help desk, understanding and executing legal contracts, searching and extracting information from a document corpus, processing receipts and invoices, orchestrating a complex supply chain, enforcing quality checks in manufacturing, or tracking revenue and expenditures so you can close the books at the end of every quarter.

In all these cases, AI drives intelligent workflows that deliver tangible business outcomes. In customer service, for example, AI-powered workflows help customers by surfacing and summarizing relevant information, delivering answers to their questions, recommending next best actions, and identifying and automating resolutions of common request patterns.

AI can help customers troubleshoot a problem, kicking off a workflow for immediate, automated resolution via self-service. AI can also route an issue to the appropriate agent while arming the agent with relevant information. It can guide troubleshooting and recommend resolutions.
As a result, service reps get to focus on what humans do best—creative thinking, customer interactions, and unpredictable work.

Enterprise AI is evolving rapidly, as companies race to digitize 20th century processes and business models. This points to a business future where machine and human intelligence work in concert to deliver experiences that empower customers and employees.