Almost Human, Near Perfect – Why Google Contact Center AI is a Trail-Blazer

Rob Scott

Almost Human, Near Perfect – Why Google Contact Center AI is a Trail-Blazer

If you’re fan of Google, and its vast repository of apps, multiple services, feature upgrades, and add-ons, this is a great time to stay online and connected with its ecosystem.

At Google’s Cloud Next Conference – held during the last week of July – the company shared details of its newest offering, and all that it is meant to achieve. Called Contact Center AI, Google is now collaborating with a number of partners (at last count, a dozen), spanning giants such as Vonage and Cisco. What’s more Google Cloud’s chief scientist, Fei-Fei Li discussed how the new tech had intersections with Google Duplex AI, an app meant for restaurant-like settings, and a perfect foil to traditional call centre models.

Now everyone knows the importance of call centres, and their ability to offer extensive help via live agents, making customer experience (CX) a more refined and organic pathway. AI could be a game-changer in this space, and it was this thought which led Google to try and create a machine-learning focused customer representative, working in-sync with Google’s Dialogflow program.

Dialogflow was launched in November last year, billed as a composite development suite for shaping conversational agents.

So, what can AI do to push customer experience to newer heights?

Simple. Using advanced tech and AI tools, high-quality customer service can be made available to customers across the board, 24/7, with zero down times – even as the solution aids live agents effectively anticipate and reply/react to customer requirements.

Google came out with a long list of new features and solutions, positioned to transform how call centres perform daily tasks. Here’s a rundown of the same:

A set of new features in the Dialogflow Enterprise Edition will let you access AI-powered virtual agents for the contact centre. These include phone-based conversational agents – entitled Interactive Voice Response (IVR). The new Google Contact Center AI solution will carry these additional Dialogflow features, along with other tools to help live agents connect intelligently with customers, while accessing heightened analytics.

Understanding what Dialogflow can implement

In the beginning, Dialogflow was envisioned as a platform helping developers build new and improved interfaces, crafted for deeper, personalised engagement. Today, the Dialogflow community has pushed beyond 600,000 developers, and is recognised as the ‘go-to’ tool for creating platform-agnostic conversational experiences, with minimal expertise in AI or natural language processing.

Over the last year, Dialogflow launched several updates. Let’s consider the elements or add-ons that could be specifically useful for contact centres. On a basic note, these are meant to automate every day interactions, allowing live agents the opportunity to focus on complex and layered issues:

Phone Gateway Beta

This helps assign an active phone number to one’s virtual agent, and calls can commence in just under a minute; Remember all of this is happening without any dependence on infrastructure, on a platform that’s geared to scale at your pace.

Knowledge Connectors Beta

These can evaluate unstructured documents such as FAQs or knowledge-driven pieces, expanding pre-built intent with automated response mechanisms. The additional data garnered from your knowledge-base is artfully combined with your Dialogflow agent, cutting down on complex integrations with disparate tools.

Automatic Spelling Correction Beta

Often natural language understanding hits a stumbling block in spelling & grammar issues, in text-based conversations. Dialogflow will automatically correct spelling errors through tech that’s largely akin to what’s at work in Google Search, and its other products.

Sentiment Analysis Beta

This is based on Cloud Natural Language API, and can offer the option to inspect a request, scoring user attitude as either positive, negative, or neutral. The score as registered can then flag scenarios where passing the baton to a live agent is the most pragmatic recourse – in cases which needing complex trouble-shooting or a more, sensitive, high-touch experience.

Text-to-Speech Beta

There’s no denying that ‘Voice’ is at the heart & soul of conversational IVR, and Dialogflow intends to offer users busy in building the same, the chance to action reinforced text-to-speech capabilities with contained development complexities. This is architected through native audio response to Dialogflow, in alignment with existing Speech-to-Text capabilities. By utilising DeepMind’s WaveNet tech, it addresses gaps in voice-quality, with human voice by 70%, and counting. And when used in tandem with the Dialogflow Phone Gateway, it’s also optimised for phone calls.

Watch Google explain exactly why Dialogflow holds immense potential for new-age contact centres, here:

Here’s how Contact Center AI can make a genuine difference

Enterprises – no matter how big or small – demand an all-encompassing solution which can effectively integrate advanced analytics with live agent support. Contact Center AI, as mentioned, is powered with all the newest Dialogflow features as listed above, working in perfect harmony with a clutch of additional tools, customised for sector-specific operations:

Agent Assist  Alpha

This can support multiple conversations, offering relevant and vital information to the agent. For instance, suggested shortcuts and articles for completing relevant tasks, will be available in real time.

Contact centre analytics (Conversational Topic Modeler)


Generally, Conversational Topic Modeler actions Google AI to assess historical chat & audio logs to unravel insights on topics and trends in customer interactions. These could significantly contribute to the gradual development of a virtual agent, specifically programmed to manage contact centre needs, and impact emerging areas such as knowledge-base content and improved agent training.

You can watch Daryush Laqab, Product Manager for Google Contact Center AI, explore all the possibilities during an exclusive session at Cloud Next 2018, here:

All’s well that ends well – why Contact Center AI is a great idea – always in motion

Thanks to Google’s wide portfolio of partnerships, the Cloud Contact Center AI is ready for integration with technologies and capabilities from other solution providers.

In fact, less than a month ago, Google was ready to start real-world testing of Duplex, under contained conditions. The company’s I/O developer conference was agog with excitement when the app was showcased as an illustration of how AI could complete simple tests, while remaining almost identical to human functionalities.

Clearly, Google is looking at AI as a multi-pronged fork – a growing tree of possibilities, each linked and yet separate from the other, from making a restaurant reservation (as in Duplex), or enhancing engagement with customers via Content Center AI.

The writing on the wall, is actually far simpler: transforming AI from the kernel of an idea, into a living, regular, common, and connected entity in our daily lives, from homes to offices, and from neighbourhood watering holes to hyper-active call centres.

Read more contact center news stories here.

Join our Weekly Newsletter