Skip to NavigationSkip to Main Content
Log In Log In using your Salesforce account

Think On It: AI in the Age of Automation

What’s the true state of customer service in an age of AI-powered automation? A recent webinar with Salesforce, DigitalGenius, and 4C covered a range of topics that sought to shed light on the next iterations in how human and machine work best together - along with how solutions from Salesforce Einstein and DigitalGenius are helping to fuel innovations in these areas.

 

Here are the key takeaways from the webinar (and feel free to watch the full session here).

 

1. DigitalGenius and Salesforce Einstein are complementary technologies

The AI space has heated up in recent years, with many rushing to deliver - or at least market - AI products and services at scale. Salesforce is no exception. Einstein AI is the company’s answer to growing volumes, complexity and scale in sales, marketing and not least in customer service.

 

While Einstein Bots was still in development, DigitalGenius often fielded questions about product differentiation, especially as DigitalGenius signed on marquee Service Cloud customers such as Air France, IATA, KLM, Zopa, and others.

 

What we learned in the webinar is that Einstein Bots are great at dealing with guided conversations. DigitalGenius’s Chief Product Officer Ivan Harris offered useful examples such as “fixing a router or how to make a pizza”, in which multi-step interactions are required to resolve the customer’s query, and which take the customer down a predefined journey to guide or diagnose a specific problem.

 

Clement Tussiot, Senior Director of Product Management at Salesforce and focused on Einstein Bots, summed it up, “if you want to automate a large chunk of simple requests in a scripted manner or using NLP, Einstein Bots is for you”.

 

DigitalGenius, alternatively, uses Conversational Process Automation (CPA) to understand the nature of the customer’s question and collects and transmits sufficient information through third-party systems in order to provide a single response to resolve the query. This is more appropriate to use cases such as “where’s my refund?”, in which a clear and accurate reply can easily solve the customer’s problem. “For something more complex, DigitalGenius is a good extension of the product we have today”, remarked Tussiot on how DigitalGenius complements Einstein.

 

These are fundamentally different conversations; DigitalGenius is purpose-built to automate replies and actions; Einstein Bots are meant to take the customer through a journey. This is all driving towards a complementary product vision, as Harris stated: “There will be some intents where Einstein bots will be more appropriate to use and others where DigitalGenius AutoPilot can resolve the case end-to-end.”

 

Depending on the customer intent, either the guided or single response journey will be launched, providing a customer experience totally in sync with their needs. It also helps push agents toward handling increasingly complex and nuanced conversations, which incidentally lead to higher job satisfaction for agents.

 

2. The elegant Einstein-DigitalGenius handover.

Joining DigitalGenius and Einstein Bots together to resolve a wide range of customer query types certainly increases the total automation volume, yet what happens once AI confidence levels drop? In cases where the AI is no longer able to address the query, or when a customer asks for a human interaction, DigitalGenius CoPilot will help agents resolve cases that are escalated from Einstein Bots, or from DigitalGenius AutoPilot.

 

Tussiot explains, “Einstein Bots can quickly do the low-hanging fruit, and in the context of the handover, the agent can be super-powered by DigitalGenius recommendations,” meaning that wherever there is ambiguity, where the AI lacks confidence, or when the customer wants an agent, DigitalGenius is on hand to help the agent resolve the problem quickly and consistently. When businesses want to automate suggested answers, they can “automate the DigitalGenius answer straight through Einstein Bots, which provides the framework for conversations”.

 

The result, Tussiot states, is “instead of covering 50 topics, you cover 200 in a quick manner.” The customer experience will remain frictionless, the customer remains able to communicate through their preferred channels, and they will benefit from several layers of AI to get them their answer.

Rita Martin, Head of AI Practice for 4C, added, “This ensures the customer does not need to repeat the same answers to the same questions”. Sounds pretty good to us, and to customer service leaders & agents worldwide.

 

3. Einstein Bots puts other chatbots to shame.

The GA release of Einstein Bots has been eagerly awaited for, and Salesforce didn’t disappoint. The bot experience is a cogent and straightforward one, whereas a typical, average chatbot falls flat.

 

Salesforce has carved out a compelling product for two reasons. First, the product approach is framed with the right expectations. Einstein Bots do not solve 100% of customer questions, and they’re not supposed to - Einstein Bots have delineated boundaries, target a clear outcome and have a stated reason to exist. (For more on the do’s and don'ts of chatbots, have a read of a recent DigitalGenius blog post).

 

Einstein Bots also stand out from other chatbots because it is a native Salesforce product. It’s already plugged into Service Cloud, collecting valuable interaction data, customer interaction feedback, and steering customers to next best actions to enhance the customer experience. This matters because deployment becomes straightforward, and businesses can safeguard a customer experience that is consistent with the way they engage with customers.

 

4. AI-driven Process Automation will deliver tremendous value in Customer Service.

DigitalGenius has the ambition to automate substantial amounts of customer service traffic. This includes not only conversations but actions.

 

The company launched Conversational Process Automation in Summer 2018, which connects conversations with the back-end processes that resolve them, but the notion of full automation has been our north star since our founding in 2014. Since then, many businesses have seen the light in this approach (including our customers), while others have followed suit and are now attempting to offer similar solutions.

 

That Salesforce aligns with our goals and has similarly targeted process automation is further validation of the impact DigitalGenius’ automation posture will have, and is having, for customer service teams. Salesforce will enable Einstein Bots to initiate actions and processes in Service Cloud in their Winter 2019 release, eventually expanding capabilities into full Lightning Flow enablement.

 

We’ve seen first-hand from numerous customers just how much of daily customer service work is not conversational, but rather process-driven: tasks such as checking information, updating data and switching between systems. Our customer experience is also validated by a recent survey by Canam Research we commissioned, one that shows that 65% of customer service teams are forced to use between one to four systems to resolve simple queries. Automating processes will fundamentally transform customer service.

 

5. Stick to the fundamentals.

DigitalGenius has progressed tremendously since launching with KLM Royal Dutch Airlines in November 2016 - as has the overall application of AI, business understanding of the technology, and approaches to data science.

 

However, some maxims haven’t yet gone out of style:


  1. Start simple and then scale out: This is a guiding principle of many AI companies, DigitalGenius included. Tussiot recommends that for any project, “Phase 1 should cover a few topics with high impact - don’t start with 500 intents”
  2. Historical data is still paramount: To launch, DigitalGenius leans heavily upon historical data, as do many AI-powered businesses. While the Einstein Bots currently has bot-builder functionality only, Tussiot stated that Salesforce will be adding historical training functionality in 2019.
  3. Expertise matters: Despite endless marketing espousing “2-clicks to AI”, deploying Enterprise-level AI solutions doesn’t happen overnight. Tussiot warns that as deployments become complex, “It’s important to bring in people who understand the technology, and who don’t get lost in the basics”. This is where DigitalGenius and 4C have considerable experience to ensure that in every project, there will be the right blend of data science, business and project management.




Phill Brougham leads product marketing at DigitalGenius, and is focused on driving business adoption of AI-powered customer service automation tools.



Most Recent Articles