Enterprises explore AI voice assistant technology

Enterprise use of voice assistants

There is a wide range of opportunities for enterprises to take advantage of voice assistants, also known as smart speakers. Intelligent voice assistants have the ability to act as additional, augmentative resources for organizations by leveraging conversational technology and AI-enabled, cloud-based processing power to perform tasks that would otherwise be assigned to administrative assistants or more basic software.

Voice assistants are gaining traction in administrative use cases, such as scheduling meetings, setting reminders and assisting with conference calls. In November 2017, Amazon released Alexa for Business which integrates with company calendar and conferencing systems to allow conference room attendees to automatically start and control meetings using their voice.

In addition, companies are looking at using voice assistants to enable employees to connect with internal departments, such as HR or IT, to facilitate internal support. Google similarly announced support for free calls to phones in the United States, signaling pending support for enterprise conference call use cases.

Enterprises are also eyeing voice assistants to provide at-desk search and queries of company resources. Rather than using company portals or connecting with web-based knowledge management systems, voice assistants can integrate with these systems to allow for voice-based queries and conversational interaction with enterprise information resources and corporate data.

Enterprises need a developer ecosystem

Since most enterprise use cases for voice assistants will be customized, it's no surprise that vendors like Amazon, Google and Microsoft are aiming to simplify and separate the process for the development of internal-only skills for enterprise use. These vendors are looking to create an enterprise-only set of skills that can be accessed by a wide range of users in the organization. These skills would not be visible or accessible by non-enterprise employees or those of competing organizations and would be managed and developed by enterprise IT and development staff.

Similarly, voice assistant manufacturers are pursuing integration within existing enterprise software and tooling, providing voice assistant capabilities on top of platforms that already exist in the enterprise, such as ERP or CRM tools. In this manner, instead of having to use web or mobile interfaces to access internal apps, voice assistants can provide a conversational interface that can simplify the interaction and empower greater use of corporate systems.

Enterprises have unique requirements for AI voice assistants

There are a number of major differences between consumer adoption and enterprise usage of voice assistants that limit the feasibility of deploying off-the-shelf devices in corporate environments.

Whereas in consumer use each device is registered to an individual with their particular profile or personal account connected to the device, enterprises need these devices to be used by a wide range of people in the organization and be tied to the corporate account, with specific license and usage level requirements. In particular, enterprises require tiered access levels for voice assistants, with some users having administrative privileges. Any charge-incurring uses of the devices may need to be turned off for corporate users with special users approving all charges, which are billed to a corporate account.

In addition, enterprises require the integration of voice assistants with corporate data and knowledge repositories, enterprise calendaring and messaging systems, and identity management infrastructure. Corporate users want to avoid an additional support headache when managing AI voice assistant devices, and as such need to make sure that the devices can be centrally provisioned and deprovisioned on the network, and users similarly centrally managed.

These integration requirements are not insignificant and, as a result, we can expect voice assistants that want to provide enterprise-level support to charge accordingly, with monthly subscription plans that are not currently charged to casual consumer users.

Challenges with enterprise adoption of voice assistants

Despite the promise of voice assistants and conversational technology in the enterprise, there are a number of significant challenges and barriers preventing widespread adoption.

From a technology perspective, the lack of voice authentication is a significant concern for enterprise IT administrators. Without the ability to know for certain who is requesting a specific skill or capability, it becomes difficult to control access levels or grant access to private information.

This means that any skills developed for voice assistants have to be for such generic information that any user can request the voice skill without security or privacy concerns. As such, this relegates the devices to use with only lower value information or scheduling-only tasks. Once two-factor authentication or other forms of secure, voice-based authentication can be put in place, more significant use of the technology will be possible.

Companies are already eyeing AI voice assistant devices cautiously and, often, distrustfully. Enterprise security officers are rightly concerned about always-on microphone snooping devices that send audio information outside the enterprise firewall.

Because, to a certain extent, voice assistant technology is a black box, there's no way to guarantee that confidential conversations, data and interactions won't be inadvertently, or even maliciously, shared with unauthorized third parties. For example, a couple in Portland, Ore., recently reported that parts of private conversations were shared unintentionally with some of their contacts by an Amazon Alexa device.

Even if enterprises implement somewhat trivial applications with voice assistants, they shouldn't allow their use in secure facilities or private spaces unless they can be unplugged or reliably disconnected when not in use. This makes the devices somewhat useless if voice-based triggering must be turned off.

At the end of the day, for these voice assistants to provide value to enterprises, they need to prove themselves to be trustworthy, valuable resources. As such, vendors hoping to penetrate and dominate the enterprise ecosystem for voice assistants need to focus on addressing key integration, application development, provisioning, security, privacy and trust issues before they can find widespread adoption and traction.