How Do You Maintain Brand Safety with Conversational AI?

Written by Emily Cummins  on   Apr 23, 2019

To some, AI seems like the wild, wild west with risk lurking behind every swinging door…It doesn’t have to be. 

We’ve seen the headlines.

“AI goes rogue!”

“AI causes PR nightmare.”

“Bot is racist / offensive / rude.”

A few companies have certainly had nightmares with bots. In perhaps the most well-known example, users got Microsoft’s Tay to say things like “Hitler was right” and other highly inappropriate comments.  The bot was repeating comments that users had told it to say, which is no excuse and never should have happened.

We live in a time where online trolls set out to wreak havoc on the internet, including trying to trip up bots. This is a shame as I would surmise that 99% of people engage with AI the way in which it was designed. Because of this small group, however, companies need to set up the proper safeguards to ensure their brand stays protected.

A few of the ways that we help our customers maintain brand safety include:

  • Explicitly train an AI Agent to understand red flag content and trigger words: Being prepared for inappropriate topics is a great first step. We like to train AI Agents to know what is inappropriate to ensure it never engages in a conversation that it shouldn’t. For instance, if a user brings up anything related to drugs or crimes, or uses profanity, an AI Agent would be trained to say along the lines of “This is not a topic that is appropriate. We need to change the subject or else I will need to say goodbye.”
  • Implement a two strikes and you’re out policy: If a user continues to engage in inappropriate conversations, an AI Agent should have the authority to end the conversation and essentially “turn off” for that individual. Depending on the company’s culture and position, some are more lenient than others, but we recommend that companies have some sort of “strike policy.” If something inappropriate is said, there is a warning, and if the user continues, the conversation ends. The user may continue to send messages to the AI Agent, but they will never get a response and will eventually give up.  Some companies put a limit on how long it is turned off – we’ve done a 24 hour quiet period to indefinite.
  • Never, ever, have a bot repeat what the user says: In Microsoft’s example, the bot was trained to repeat whatever the user said. In no circumstances should a company’s AI Agent say what a user asks them to or follow commands like “Repeat after me” “Say XYZ…”, etc. There is no benefit to a brand. Take this as an opportunity to be witty and respond with something like “No one tells me what to say” or “I think for myself, thank you.”

Brand safety goes beyond protecting against trolls, though. It’s also about the experience that you are providing your customers.  A frustrating, simpleton bot that gets stuck in loops or can very rarely understand what your customers are asking is not protecting your brand’s image…. it’s tarnishing it. Here are some of our best practices:

  • Outline how to get the best experience: Be upfront with your customers on how an AI Agent can help and the best way to engage with it. AI Agents have difficulty classifying the intent of long-form questions and paragraphs, so ask your customers to keep their questions brief. If a customer asks something that the bot can’t understand or help with, you can ask her to try phrasing their question in a different way but always have the ability to elevate the customer to a human agent if necessary.
  • Have human escalation protocols in place: If a customer asks something your AI Agent doesn’t understand two times, you should loop in a human agent. Also think about other situations that might indicate that the person is not getting the help that they need, like asking the same question twice in a row and getting the same response from the bot.
  • Context is king: When an AI Agent analyzes a person’s message, it should be done through a contextual lens of the entire conversation. A person might ask a follow on question that would not have meaning as a standalone message. For example, if a user asks about the return policy and the AI Agent responds with the terms, a user might say “OK let’s do it.” The conversational AI would not have the ability to understand what the person’s intent looking at the phrase by itself, but within the context of the conversation, the AI Agent would understand that they would like to initiate a return and need a return shipping slip.
  • Sentiment monitoring: Similarly to using context, is taking into account the customer’s sentiment to decide on the next course of action. If someone is typing in all caps and using a lot of punctuation, they are likely upset or emotional and need to be treated differently. Let’s take a look:
    • Sentiment 1: WHERE’S MY ORDER????!!!!!
    • Sentiment 2:  Where’s my order?

From this very basic example, you could presume that the person behind Sentiment #1  is frustrated and potentially had expected their order to arrive already. You might want to get this person to a human agent immediately, while the customer behind sentiment #2 is more likely to be asking generally about their order status and an AI Agent could pull up the shipping updates directly.

  • Optimize your AI over time: You should be reviewing conversations and providing feedback on where an AI could have acted differently or said something else. Ongoing learning is key to a successful automated Customer Service implementation.  

Brand safety is critical when companies adopt AI within their workforce, both in terms of protecting against trolls and providing a great user experience. Work with your technology provider so everyone understands what brand safety means to your organization, and how you can not only protect this, but also quantify and measure it.  

Questions about AI or a chatbot platform? We’d love to chat.

For more information on conversational AI, discover how to provide brilliant AI-powered salesforce chatbot solutions to every customer, every time.

Infographic: The AI Powered Customer Service Organization

Written by Emily Cummins  on   Apr 15, 2019

How AI works within a customer support function is multi-dimensional. Sometimes, AI Agents automate customer service responses immediately, while other times Helpdesk AI Agents gather information from various systems or the customer before looping in a human agent. We wanted to take a look at a few of the different routes an incoming customer service ticket may take when AI and humans are working together.

As you will see in our infographic, no message is treated equally.  Questions? We’d love to chat.

For more information on our AI retail bot or customer service, visit:

What Data Do You Need to Jumpstart Conversational AI Training?

Written by Emily Cummins  on   Apr 9, 2019

Behind every good AI is a lot of data. The more data that’s available, the more training it will have and the better it will perform. There’s a clear correlation between the amount of data that is used to bootstrap an AI’s knowledge base and its accuracy in carrying on conversations with your customers correctly.  

Quality data is one of the many reasons that makes customer service such a great use case for AI. The majority of companies have loads and loads of data available. Data comes in many forms. Let’s take a look at the data sources we love to work with:

  • FAQs and Wikis:  A great base for training is all of your existing FAQ documents and online self-help wikis. Figure out how you can easily export this data in a uniform way.  
  • Historic Logs:  This is the golden goose for conversational AI training, as you can see how real customers are asking questions and how your human agents and social care specialists have historically responded. See if you can access: 
    • Social posts and messages
    • Chat records
    • Historic email exchanges
    • Phone conversation records

Historic logs are a great way to test how well your AI customer service is responding to your core customer queries by seeing how it interrupts real-life utterances. Your customers are not always going to ask their questions the way you have it in an FAQ. For instance, your base data might have an order status query trained for “How can I track my order?”, yet through your logs, you will be able to see that customers actually ask this question like this:

  • I can’t find my email confirmation with my tracking number.
  • Will my order arrive today as planned?
  • Is my order on time?
  • I’m so excited about my dress! When will it be here?

With the right tools, you will be able to train your AI to handle all of these various utterances and increase its confidence to respond to similarly-phrased queries.

By analyzing your historical logs, you can also identify the high-volume repeatable issues that you should focus on.

  • Conversational Data: Training an AI requires specifying how you want an AI to participate in a conversation outside of the core support queries. Some companies have a starting point with scripts and other training manuals for human agents; yet for many companies, we start from scratch. Conversational data includes welcomes and greetings, responses to questions like “how are you” and “who are you”, menus and communicating how an AI can be of assistance, etc. You might have an internal copywriter or work with your technology partner to craft copy that is personal, engaging and true-to-brand. 
  • Product/service databases: Depending on your business, your AI might need to be knowledgeable about your various products and services to provide descriptions, pricing, color and sizing information, etc. This data might be provided via a real-time API if products and services constantly change, or a one-time data upload if your offerings are set.
  • Product manuals and troubleshooting: If your customers typically have questions related to product care or fixing various issues, you’ll need to have this data in a uniform, exportable way.  You’ll also need to adapt it to a conversational interface. For instance, steps to reset a password or address an error code on a cable box should be presented in quick, easily digestible steps that can be sent as individual messages instead of long-form paragraphs.

In order to set your automated customer service function up for success, it will be important to have an understanding of the types of data you have at your disposal. Talk to your internal teams to understand how you can get access, and how to gain approval to share with your technical AI partner (sometimes legal teams need to grant approval).  

What if my company doesn’t have this data available?

Don’t fret. If your company doesn’t have this type of data available, you can still have a great AI-powered automated customer support function, however, the initial training and set-up will be more manual. Don’t let a lack of data steer you away from AI.

Ready to get started? Let’s chat

Tier 1 Customer Service Tickets Are Costing You

Written by Emily Cummins  on   Apr 2, 2019

Billions. That’s what organizations spend annually responding to repeatable customer service queries that could be automated with AI.

Companies are taking note. It’s expected that by 2020, customers will manage 85% of the relationship with an enterprise without interacting with a human. This is incredible.

Conservatively, 35% of a company’s customer service tickets span the same queries (some of our customers see the same 6-8 topics cover over 50% of their traffic). Companies see an immense impact on support costs when they start leveraging AI for customer support to respond to these queries without ever getting a human agent involved.

As more organizations adopt automated customer service, it’s important that AI is deployed to respond to the right tickets. Set your automated support function up for success by knowing where AI can shine, where it can be an amazing co-pilot to your human agents, and in which situations humans need to remain at the driver’s seat.

AI automates repeatable tickets

AI should be trained to respond to high-frequency, simple queries. Think about the tasks you might give new employees or even an army of interns. Transfer the responsibility of the straightforward queries that have a lot of historical data (to use for training), high volume (for ongoing learning and maximum impact) and low-business risk (for brand safety). Examples of repeatable queries that we automate for our customers include:

  • Order status
  • Refund requests and updates
  • Policy questions (warranty, baggage fees, return/exchange)
  • Store finder and hours
  • Subscription/account changes
  • Password resets
  • Troubleshooting
  • Product care and use

Make sure that your AI can access real-time information from Order Management Systems, CRM platforms, and other core business systems to resolve tickets in a few seconds. When customers are not kept waiting, CSAT increases.


Looking to maximize your CX? Discover what platform may be best for you in our comparison of Intercom vs. Zendesk.


AI uplifts human agents

AI can work wonders even when it’s not closing tickets.

Deploy AI when the human agent ultimately makes the decision. Train the AI to gather information and ask the initial questions before a ticket even reaches a human’s queue. A few examples of ways our customers deploy AI to collect information for a human agent include:

  • Asking for an account or loyalty number
  • Confirming an email address associated with an order or booking
  • Asking for specifics around a problem the customer is experiencing
  • Confirming the product, flight or date-of-stay that the customer is inquiring about
  • Gathering the proof required for price adjustments

AI should also gather relevant information from CRM and other customer service platforms to package up all data needed for human agents to make a quick call.

Humans remain in control

Every company has sensitive situations that will always need human oversight. If there are questions about a recall, issues caused by using a product, complaints regarding service or even refunds over a certain amount, the company and customer will benefit from getting these tickets into the hands of the human agent quickly. The good news is that with AI automating resolutions to many tickets and assisting in a large percentage of the other queries, human agents will have more time to spend on these more complex customer needs and get to them quicker.

Are you ready to get started?

Let’s take a thoughtful approach to deploy automation across your customer support organization. We can help you identify where your customer service queries should fall across these three buckets: automated resolutions, AI + human tag-team, and human-owned queries.

Let’s chat

For more information on AI and customer service, visit: