Is Solve the right fit for your customers?
Your team receives repetitive one-touch type questions and systematically applies the same response over and over again. If your team could benefit from having these types of tickets removed from their queues, we can help!
How does Solve for Email work?
Forethought will identify common and repetitive customer questions associated with macros or knowledge articles from your team. We'll use artificial intelligence to learn from your past tickets and automate this process. Then, when a customer ticket comes in, Forethought will read, understand, and respond to it with the relevant macro.
How does Solve Widget work?
The Solve Widget is an interactive, customizable widget that lives on your organization's webpage. It directs customers to their best solution and can surface the best-fitting article from your knowledge base to answer their issues or create a ticket for your team to answer. As the widget's owner, you will have access to editing this widget through the Forethought dashboard to continue improving the experience and the number of questions it deflects.
Unlike most Customer Support AIs, we use Natural Language Understanding (NLU), a cutting-edge AI technology, to train on your specific data. Our use of NLU means we are able to launch Solve in as little as 3 business days with above 90% accuracy. That, of course, is not the case for everyone. For more information, read some of our case studies here.
What kind of workflows can I use for Solve Widget?
Intents are essentially classifications for specific problem types. Solve tends to work best for repetitive support cases with rigid questions and answers.
Some examples:
- Password Reset / Login Assistance
- Pricing Inquiries / Quotes
- Delivery / Shipping
How many phrases are needed to train a Solve Widget Workflow?
An intent refers to the goal your customer has in mind when typing in a question or comment to the widget. Intents are how Forethought’s AI interprets what your customers are asking and determines the appropriate response to return, which deflects support tickets from being created and reduces your agent's workload. Solve Widget works best for frequently asked questions from customers that can be resolved with text responses that provide additional information or with a knowledge article.
You will need a minimum of 2 training phrases for each workflow! However, we recommend adding more than 2 to train the model on a wider variety of different ways a customer may ask a question and to differentiate intents.
How much data do I need to have?
Our customers see immediate value when our Solve products have at least 1,000+ tickets per week. This is because there must be a sufficient historical usage of the macros you would like to Solve Email to use in response to your customers. Alternatively, if you do not use macros and prefer to send knowledge articles to your customers, having articles that address frequently asked questions is needed in order to allow Solve Email to send the appropriate article in its response to your customers. This allows us to ensure there is sufficient data to train and create a custom AI model that will accurately send the expected macro or knowledge article response back to your customers.
For Solve Widget, you don't need any data at all! However, utilizing a knowledge base to deflect questions or utilize our other integrations is the best way to get value from our product.
What type of actions can it automate?
Within the Solve Widget, you can perform actions to pull, update, create, or delete information by using any valid and public API endpoints. You can customize these actions to automate and deflect inquiries from your agents. For more details, read about our widget action build feature here.
What platforms do you support?
We support various platforms, such as Zendesk, Freshdesk, Salesforce, Airtable, Intercom, and many more. For more information, see Available Connectors and their Requirements.
What languages does Solve Widget support?
For more information, see Supported Languages in Solve.