Insight

"How may A.I. assist you?"

Conversational AI agents can boost employee performance, productivity and outcomes

One of the most visible applications of AI are “conversational agents”—chatbots and intelligent assistants that interact with people via voice or text channels, on devices such as smartphones, automotive infotainment consoles, and smart speakers. Having conversations with AI is becoming routine for consumers— and soon it will be for employees, too. In the workplace, conversational agents can help workers interact more seamlessly with each other, streamline office operations, execute internal process, and deliver data more efficiently.

Classic employee-facing technology initiatives don’t exist in a vacuum, and neither will conversational agents. Their development and deployment—and ultimately, their capabilities—will be driven by employee expectations about what they should do, and those will in turn be shaped by the external environment.

This paper was created for those individuals who are considering embedding conversational agents inside their organization to streamline and enhance internal operations.

This paper includes insights on:

  • Where and why to use conversational agents
  • How to drive enterprise adoption of conversational agents
  • The building blocks of successful conversational agent deployment
  • How to grow conversational agent capabilities as technology evolves

Download: "How may A.I. assist you?"

Conversational AI agents can boost employee performance, productivity and outcomes

 

Conversational agents: The next frontier of business productivity

In the past, an employee with a question about how to get something done would ask a knowledgeable colleague for an answer. In the future, they’ll likely ask a conversational agent, and artificial intelligence (AI) will answer their question.

AI is racing out of the lab and onto the front lines of business, driving companies around the globe to reinvent how they operate. AI has become essential to competing in today’s marketplace—for predicting what internal and external customers want and discovering how to serve them with more rewarding and personalized experiences.

Where and why to use conversational agents

While customer‑facing conversational agents have grabbed attention, leading companies are interested in how bots can streamline and enhance internal operations. Automation has been a lever for business process improvement for some time, but the advent of conversational agents now makes automation naturally accessible to workers as part of their day to day.

Conversational agents can provide the most significant benefit in three areas:

Assisting the team: The conversational agent—trained to observe and capture team interactions—accumulates important institutional knowledge. As the bot learns more about the group’s patterns and activities, it becomes like a virtual team member, with a photographic memory for past discussions, action items, tasks, and reminders. This type of bot is often deployed on collaboration platforms. For example, conversational agents on platforms like Slack, known as “Slackbots,” are widely in use as office assistants to complete tasks, such as scheduling meetings, translating text, and ordering lunch.

Streamlining office operations: A task-oriented conversational agent can enable self-service of administrative tasks, alleviating office tedium and making it faster and easier to complete basic work. A classic example is the IT helpdesk bot, which leads employees through common multistep procedures, such as password resets, without human intervention. More sophisticated intelligent assistants allow users to book meeting rooms or guide employees on content searches. Conversational agents also provide a cost-effective way to thoroughly engage sales leads because they can understand intent, sentiment, and urgency and have shown the ability to set 20–30 percent more appointments than human teams working alone.

Delivering information: This type of conversational agent provides on-demand information and guidance on matters important to the team. The bot alleviates the pain of scaling response teams to address increased demand for company data. For example, a “financial reporting bot” can provide granular fiscal figures at month-close, reducing the demand on financial analysts to craft custom reports. As a portal for vast amounts of information, this type of bot also allows faster ramp-up and repurposing of resources between areas, thus reducing retraining.

Driving enterprise adoption of conversational agents

To keep pace with AI market leaders, enterprise conversational agent adopters should take a page from their book. Four capabilities stand out as foundational to all conversational agents deployed in the workplace:

  1. Conversational agents must learn.
  2. Conversational agents must fail usefully.
  3. Conversational agents must be personalized and provide assistance that not only meets employee needs but also meets them in a familiar way.
  4. Conversational agents must prioritize user experience.

These four capabilities help ensure that the conversational agent finds adoption within the organization and sustainably delivers return on the investment put into its development. Tying adoption metrics back to the initial business case for deploying the conversational agent substantiates the impact that the agent has on overall operations.

Developing building blocks of successful conversational agent deployment

To ensure conversational agents deployed in the workplace deliver value and can be sustained, organizations must master three key areas: technology strategy, organizational capabilities, and organizational culture.

Technology strategy

Deploying conversational agents requires a clear understanding of the technology underpinning the bot interface, task fulfillment, and knowledge curation and governance.

  • Bot interface: The bot interface defines how employees interact with the conversational agent and is the key driver for fostering adoption. Usability and user experience should be assessed, defined, and validated as early as possible in the development process.
  • Task fulfillment: Task fulfillment is how the bot interfaces with other systems to take in information and initiate processes. The bot must be able to consult existing sources of truth, acting as a portal to existing sources of record.
  • Knowledge curation and governance: As the conversational agent learns and adapts, the organization must curate and organize the data it is trained on, the responses it provides, and how its persona is tuned.

Organizational capabilities

Analysts forecast that conversational agents will disrupt how organizations work as well as the kinds of work people do on the job. Offerings in the technology marketplace have already started to examine how conversational agents can improve business functions ranging from sales screening to recruiting to scheduling meetings and resources. At KPMG, we believe the range of solutions accessible by conversational interfaces will only continue to grow across the array of enterprise functions and verticals. Organizations will need to onboard new skill sets to support the deployment of conversational agents. They will also need to redefine operations to get the most value from human employees and bots based on their distinct capabilities. Let’s examine some of the specific roles both technical and nontechnical people will play—and skill sets needed—to enable enterprise adoption of conversational agents:

  • Subject matter experts from the areas where conversational agents will be deployed and will document the context surrounding their day-to-day task completion: the language they use, the procedures they follow, the data they leverage, and how they measure success.
  • User experience designers, data scientists, and data engineers will develop processes for logging data from subject matter experts and, in the future, bot users.
  • Ontology managers and knowledge management experts will build a system that not only speaks the language of the organization but also can tie words and phrases to meaningful entities and create context within the organization.
  • Communications staff members will write task-oriented dialogue flows to define how the conversational agent will converse and interact with users. These scripts must cover the multiple different paths a user might take to reach an end goal.

Organizational culture

Bringing on bot employees is a major change for any workplace—and it’s rarely without opponents. There are those who think technology can’t get the job done and those who worry it will eliminate their role. In such a strained environment, how can humans and bots hope to work together effectively? Change management and governance are crucial but often overlooked areas of implementing conversational agents. They are essential in creating a culture where conversational agents can flourish alongside humans.

Interestingly, the bot can actually do a lot of the change management and governance work itself, driving long-term adoption and ROI.
 

KPMG’s approach to deploying conversational agents at work

This approach has helped numerous organizations–from major retail banks, to mortgage lenders, healthcare payers, insurance and investment firms–successfully leverage conversational agents to improve internal operations and deliver information and data more efficiently and effectively.


1. Define conventional AI strategy

  • Identify where and why a conversational interface could be used  within existing organizational roles and processes
  • Define platform strategy and guiding principals
  • Define business value metrics aligned to organization strategy, and tied to usage

2. Assess current capabilities

  • Profile any existing use of conversational AI within an organization
  • Assess supporting data science and technical roles
  • Inventory available knowledge bases and data sources

3. Use-case focused design

  • Create personas for both users and AI agents
  • Design user experience: voice, text, multimodal, conversation medium
  • Feasibility testing campaign with preliminary conversation designs

4. Minimum viable product (MVP) creation

  • Prioritize ‘core’ capabilities and features to realize useful state
  • Begin user experience and adoption tracking, and optimize design choices on metrics
  • Hone human-in-the-loop training and design roles to illustrate how subject matter experts will provide feedback to the bot

5. Iterative learning and growth

  • Push learning strategy to feature growth initiatives
  • Capture user dialogues and feedback, and exercise next-best-feature discovery
  • Sustain AI feature alignment through bot governance and curation
  • Change management and training to drive usage adoption and continued improvement

Related content