You’ve Got Questions; We’ve Got Answers

Forethought will make your customer support team better. Before you get going, you might have some questions. And we want to provide answers. 

What kinds of things can AI help customer support with the best?

AI for customer support is no longer limited to a bot experience. AI can be used for auto-routing issues to the right place for resolution, for finding the biggest, most high-priority tickets, or for removing spam from your queue. AI can also be used on the backend as well as on the frontend. At Forethought, we believe that your institutional knowledge is a treasure trove that you should be capitalizing on! AI can empower agents to surface better knowledge faster or to respond to email tickets using macros or articles. And AI can still work on your website to help customers self-serve.


If I want to deflect tickets, do I need to use a chatbot?

There are a lot of bad chatbots in the world, and some of them will create a terrible customer experience. At Forethought, we will help you deflect needs and leverage self-service mechanisms as we see this as a valuable tool that can be really helpful to customers and agents when done well. 

If you are thinking about ways to leverage “deflection tools,” we should talk.


Is Agatha a chatbot?

The word “chatbot” doesn’t really do Agatha justice. Agatha is reading questions comprehensively, looking for homonyms, context clues, and other info that will help her answer questions more carefully and correctly. Additionally, Agatha doesn’t require the same level of implementation to get going. She ingests any historical data specific to your organization immediately and starts to help resolve issues in a matter of days instead of weeks. Agatha also learns over time so that even when business needs change, Agatha won’t get obsolete.


What makes Forethought different from other AI customer support tools?

Too frequently, using AI tools can come at the cost of a quality experience for customers. 

With Forethought, you don’t have to make this trade-off. We fine-tune Agatha so that you can answer more questions correctly without turning into a chatbot that will burn you. 

Another problem that occurs with most AI out there is that it’s rule-based. But maintaining a set of rules is as hard as you might imagine. Customer support teams end up managing all of this and end up using up the time they saved in the first place by automating things just to work out the kinks and to catch the rule-based bot’s mistakes. 

Agatha is trained on your institutional knowledge and built using machine learning, which means she doesn’t have to be taught a series of rules. She learns to recognize patterns better than a human can. Welcome to saving time and accurately solving problems using AI.


What is your pricing? 

For information on our various product offerings, go to our pricing page. Our pricing varies based on the volume of tickets and agents who would leverage Agatha.


What are you using to teach your AI?

Forethought’s AI uses past tickets, knowledge bases, articles, macros, templates, and lots of other data sources to train and become a powerful, insightful, and expert support agent overnight.


Can AI help deflect issues, even if I’m a B2B company?

Maybe you’re thinking that AI only works for B2C organizations because you’re thinking of individual customers asking a bot the same questions again and again. These kinds of traditional bots struggle with B2B support issues because it is hard to configure them to cover the myriad of types of questions that people can ask. Additionally, B2B customer issues tend to evolve over time, and answers are sprinkled across knowledge bases, making it more difficult for a decision-tree-based bot to succeed long-term with B2B issues. 

Agatha’s text-based approach is a strong choice for B2B organizations that are navigating all these different issues with content and question-type. Agatha understands the nuance of vocabulary used across other customers and their various complex queries. As a result, Agatha solves more without human intervention. If you have complex issues, Agatha’s here to help.


How do we help other members of your team get up to speed with AI?

AI is everywhere these days, and maybe you’ve gotten pushback or other teams have pushed you in your organization about using AI to make life better (or harder). At Forethought, we seek to bring together whoever within your customer support organization who would profit from implementing AI. Whether you have designated routers, a team of agents, managers who are up to their ears in tickets, or a few VPs or C-Suite folks who have questions, we’re happy to talk to them. What AI can do for customer support depends on the company, the maturity of data available in the organization, and whether the internal infrastructure is text-based enough to use Agatha’s tools. 


What does Natural Language Understanding mean?

Natural Language Understanding (NLU) is a type of AI. A lot of AI does some level of natural language processing (NLP), which is essentially matching keywords or associations within the text to help the AI gain accuracy and create effectiveness. But Natural Language Understanding (NLU) works even harder than NLP to ensure that customer queries and your internal data are understood even better. NLU uses more sophisticated models and language processing than most AI that say they use NLP. It may seem like we’re splitting hairs, but the end result is that your customers often get more and better right answers from AI the first time than they do with less sophisticated AI. 


What system do I have to use to get your AI to work? 

Forethought’s Agatha AI integrates with Salesforce, Zendesk, and ServiceNow. But we’re continually adding to our list of integrations and are happy to discuss the possibility of integrating into new helpdesks to help you better. 


Does Agatha just suggest articles?

If possible, Agatha will point your attention to the exact portion of text that is relevant to resolve an issue, either by highlighting it or calling it out in some way. This is the power that Agatha brings because she is doing more than tagging or keyword association.


What does implementation usually look like for AI? 

Implementation can end up being the most painful part of a lot of AI solutions out there. They frequently require a lot of input from customer support teams and a lot of upkeep to maintain relevance over time. Especially when you are a fast-growing company, you have no time for clunky AI. 

However, implementation for Agatha is surprisingly simple. Unlike a lot of AI and bots, Agatha starts helping the moment she starts ingesting your data. More importantly, Agatha doesn’t require major overhauls or a ton of upkeep to keep her relevant because she’s built on your data and can retrain whenever needed.


How can I protect my customers’ experience while tapping into the power of automation?

We don’t want your customers to feel automated. Most humans don’t appreciate feeling “deflected.” But self-service options can be effective, and, if automation can generate the right answer, it makes sense both in the eyes of support leaders and for the customer. 


How many tickets does Agatha need to work well?

Agatha needs about 20,000 historical tickets to read to really make Agatha work well. If your company gets about 2,000 email or web form tickets a month, she should work without any issues. 


Does Agatha learn customer-facing or internal-facing knowledge?

Agatha can learn both customer-facing data as well as internal-facing data, depending on the Agatha product that you leverage. She can differentiate depending on the customer needs and problems you’re trying to solve.


What knowledge sources can Agatha use?

Because Agatha is built on your institutional knowledge instead of a series of rules, it’s important to share what types of customer support knowledge she can ingest! She can learn customer-facing knowledge, such as help articles, cloud docs, and macro templates that companies have manually created as “sources of truth.” Agatha can also learn other, more unconventional sources of knowledge, including past customer service tickets and email conversations containing collective knowledge that is organically generated by your organization every day. The knowledge is more real-time and relevant. In the past, this kind of knowledge has tripped up AI. But they don’t trip up Agatha!


What does implementation of Agatha look like for most customers?

Agatha can index your entire corporate knowledge in a few days. In contrast, products that require specially-created knowledge can take weeks or months to set up. Agatha can also be trained in a matter of days. This because Agatha is trained using automatic learning (automatic pattern recognition from the data). In contrast, many products that require rules-based, bot learning can take weeks, months, and years to get going. Finally, Agatha integrates right into your workflow and your helpdesk, which increases the chances that your team will start reaping the benefits and seeing the benefits of Agatha sooner. 


What is Agatha’s AI technology?

Agatha is powered by artificial intelligence, but more specifically by machine learning. More specifically, she’s powered by Natural Language Processing. But even more specifically, she’s powered by Natural Language Understanding!  

Agatha is build on AI based on NLU and NLP.

Whether you’re in the market for a chatbot or an answer recommendation engine for customer support, a best-in-class AI tool for answering questions should be able to interpret the underlying intent of a question without requiring a set of keywords. NLU is a technology that interprets the “intent” of a question without keyword matching.

To better understand “intent, take the example of how Google Search has evolved over the years from keyword-based search to intent-based search. Not long ago, if someone in San Francisco wanted to know the weather, they had to search something similar to “weather San Francisco”

Now, if someone searches “how hot is it outside,” Google will understand the intent of the question even though the search never specified “weather” as a keyword.