【Google Cloud Next18】Google用AI改造企業客服中心，讓AI客服可以自然語言理解回應、還能判斷情緒

【Google Cloud Next18】Google用AI改造企業客服中心，讓AI客服可以自然語言理解回應、還能判斷情緒

News from: iThome & Google Cloud.

Google這次為Dialogflow新增的功能，訴求開發人員能以最少的工作，以自動化為客服中心達到更好的客戶服務品質。

Google宣布在其客服中心解決方案加入人工智慧新功能，這項更新包括Dialogflow企業版本的新功能，企業可以使用人工智慧虛擬代理人，進行互動式語音回應（Interactive Voice Response ，IVR），在Google雲端客服中心人工智慧解決方案，也新增了許多新功能，協助企業簡化複雜客服工作。

Dialogflow 是一個端對端開發套件，可以讓開發人員用來建置友善的網站、行動應用程式、熱門通訊平臺和物聯網裝置的對話式介面，目前社群內已經擁有60萬開發人員，過去一年，Dialogflow持續新增了許多功能，包括內嵌程式碼編輯器、多語言支持、分析儀表板、語音整合、Dialogflow企業版本和v2 API。而現在也推出了一系列功能測試版，目標為複雜的客服問題，提供自動化代理人解決方案。




Dialogflow大量加入人工智慧的功能，來提供高品質的客服服務，像是Dialogflow推出了Phone Gateway功能，現在企業可以為虛擬代理人分配一個專用工作電話號碼，在不需要新增任何基礎設施下，就能開始接聽電話，系統會自動管理電話基礎設備，進行語音辨識、語音合成、自然語言理解與編排協作，這整套的功能不需要開發人員動手拼湊不同服務，才能建構出一個完整的電話應用程式。

而Dialogflow知識連結器（Knowledge Connectors）則可以理解非結構化的文件，像是問答集或是基礎知識文章等，透過內部文件的補充，系統能夠以很少的額外工作，充實自動回應的對話體驗。Google提到，這些從文件萃取出來的訊息，可以無縫新增到代理人功能中，不需要透過其他工具，進行繁複的整合工作。

在互動語音回答中，文字轉語音是很重要的功能，Google提到，這也是他們為Dialogflow增加原生語音回應的原因，這個能力額外補充了語音轉文字的功能，雖然目前也仍是測試版，但Dialogflow的用戶現在已經能輕易的採用這個功能。

除此之外，Dialogflow還加入了情緒分析功能，可以快速的推測用戶的態度為正面、負面還是中立。還有自動拼寫訂正功能，Google表示，由於在文字的對話中，自然語言理解有時候會受拼寫與語法錯誤影響，Dialogflow現在加入類似Google搜尋用來自動訂正拼寫錯誤的功能，增強對話品質。

----------------------------------------------------------------------

Whether providing useful self-service tools, or offering detailed help through live agents, call centers make effortless customer experience one of their highest priorities. And AI can help.

AI technology and tools today have the ability to make high quality customer service accessible to all customers, 24/7, and without wait times, all while helping live agents better anticipate and respond to customer needs.

Today we’re announcing new features and solutions that will greatly benefit contact centers.  These include:

• New features in Dialogflow Enterprise Edition to help you build AI-powered virtual agents for the contact center, including phone-based conversational agents known as interactive voice response (IVR).
• Google Cloud Contact Center AI solution that includes these new Dialogflow features alongside other tools to assist live agents and perform analytics.

Empowering the contact center with Dialogflow

Dialogflow was created to help developers build interfaces that offer engaging, personal interactions. And with a community that’s now surpassed more than 600,000 developers, it’s become a trusted tool for building natural, cross-platform conversational experiences without needing expertise in natural language processing or AI.

In the past year Dialogflow has announced a number of updates, including an inline code editor, multi-lingual support, an analytics dashboard, speech integration, the general availability of Dialogflow Enterprise Edition and the v2 API, and support for multiple versions/environments.

Today we’re introducing new features that are particularly useful for contact centers. These features can automate simple, common interactions, freeing up live agents to focus on more complex issues.

Phone Gateway Beta

With Dialogflow Phone Gateway, you can assign a working phone number to your virtual agent and begin taking calls in less than a minute—all without infrastructure, on a platform that scales as much or as little as you need. All of the telephony infrastructure, speech recognition, speech synthesis, natural language understanding and orchestration are managed for you automatically so you won’t need to stitch together multiple services to build a phone application.

Knowledge Connectors Beta

Dialogflow Knowledge Connectors can understand unstructured documents like FAQs or knowledge base articles, complementing your pre-built intents with automated responses sourced from internal document collections, enriching the conversational experience with little extra effort. The added information extracted from your knowledge is seamlessly blended into your Dialogflow agent, saving the need for complicated integrations between multiple tools. Knowledge Connectors are available to try today, with unlimited usage available in the next few weeks through a new Enterprise Edition Plus plan.

Automatic Spelling Correction Beta

Natural language understanding can sometimes be challenged by spelling and grammar errors in a text-based conversation. Dialogflow can now automatically correct spelling mistakes using technology similar to what’s used in Google Search and other products.

Sentiment Analysis Beta

Sentiment Analysis relies on the Cloud Natural Language API to optionally inspect a request and score a user's attitude as positive, negative, or neutral. That score can flag situations where handover to a human agent is desirable, for troubleshooting a complex issue or otherwise offering a high-touch experience.

Text-to-Speech Beta

Voice is the soul of conversational IVR, so it’s important to give Dialogflow users building those apps robust text-to-speech capability that requires minimal development. That’s why we’ve added native audio response to Dialogflow, complementing existing Speech-to-Text capability. By taking advantage of DeepMind’s WaveNet technology, it closes the voice-quality gap with human voice (based on the Mean Opinion Score for voice quality) by over 70 percent. And when used in conjunction with the Dialogflow Phone Gateway, it’s optimized for phone calls.

Note that Google has several technologies that help users take action or communicate over the phone, and while Dialogflow and the recently announced Duplex share some underlying components, they have distinct technology stacks overall.

Enhancing the contact center experience with Contact Center AI

Giving developers the tools to build their own conversational interfaces is only part of the story when it comes to transforming the contact center. Many enterprises need a more comprehensive solution which seamlessly integrates automation with live agent support and robust analytics. To meet this need, we’re introducing Contact Center AI, a new solution that includes all the newest Dialogflow features with additional tools tailored towards contact center support and operations.

Agent Assist Alpha

Agent Assist system supports the conversation and provides the agent with relevant information—for example, suggested articles and shortcuts for fulfilling relevant tasks—in real time.

Contact center analytics 

(Conversational Topic Modeler) Alpha

Conversational Topic Modeler uses Google AI to analyze historical audio and chat logs to uncover insights about topics and trends in customer interactions. These topics and trends can also contribute to the development of a virtual agent that is tailored to the contact center’s requirements, as well as surfacing areas to improve agent training and knowledge base content.

Thanks to our strong partner relationships, Cloud Contact Center AI easily integrates with technologies and services from other trusted contact center providers.

• Platform Partners:
  • Appian
  • Cisco
  • Five9
  • Genesys
  • Mitel
  • RingCentral
  • Twilio
  • Upwire
  • Vonage
• Appian
• Cisco
• Five9
• Genesys
• Mitel
• RingCentral
• Twilio
• Upwire
• Vonage
• Tools Partners:
  • Chatbase
  • UiPath
• Chatbase
• UiPath
• Integration Partners:
  • Deloitte
  • KPMG
  • Quantiphi
• Deloitte
• KPMG
• Quantiphi

We’re also working with several of these partners—including Appian, Chatbase, Cisco, Five9, Genesys, Mitel, Quantiphi, RingCentral, Twilio, UiPath, Upwire, and Vonage—to engage with us around the responsible use of Cloud AI. This includes best practices such as disclosing when customers are talking to a bot, and education around issues such as unconscious bias and the future of work. We want to make sure we’re using technology in ways employees and users will find fair, empowering, and worthy of their trust.

What our customers are saying

How Ticketmaster is using Dialogflow

Ticketmaster is exploring using Dialogflow in a number of different ways, like integrating it with its customer service IVRs to assist with intent recognition. It’s also building enhanced self-service features using Knowledge Connectors to help address its substantial volume of inbound calls.

According to Ticketmaster product manager Tariq El-Khatib, “Today, a lot of our customer calls are ‘general’ questions about specific events and venues that have to be handled by a live agent because the details vary by event. Now, with Knowledge Connectors, we can begin to explore answering those questions within the IVR, giving our customers easy, 24/7 access to that information without having to constantly build new intents. We're excited that Dialogflow continues to align with the Ticketmaster mission to build innovative products that help provide the best services and live event experiences in the industry for our clients and fans.”

How American Financing is using Cloud Contact Center AI

American Financing is using Cloud Contact Center AI to intelligently route calls. “We’ve been a Vonage customer for three years, leveraging their unified communications and contact center solutions,” said Duane Kibbee, Director of Technology for American Financing. “Now, using the Vonage CX Enablement Suite integrated with Google Cloud Contact Center AI, we’re taking our contact center to the next level. This solution will allow us to route calls to the most skilled agents and provide after-hours support without using a third party. We expect it to create a seamless, more meaningful customer experience through contextual communications, while also reducing costs.”

Getting started

Developers can get started with Dialogflow today—visit the Dialogflow website for more information.

Contact Center AI is available exclusively through our partners. For more information, sign up here.