A small group of employees can make a large difference for AI deployments. Establishing a Cognitive Center of Excellence can help you improve how you train and refine AI. Read this article to learn how to set it up and get the most value from your implementation.
Artificial Intelligence (AI) solutions require training and customization in order to provide businesses with maximum benefit. AI systems must integrate with all front- and back-end data, and also master business processes, industry terminology, and the measured reactions necessary to show empathy through various interfaces. For IPsoft clients, this process requires what we call a Cognitive Center of Excellence (CCOE). The CCOE combines IPsoft experts and technology with our clients’ experts, back-office workers, and existing systems to ensure a smooth implementation. (Read more about this subject here and here.)
The Opening Stages
The CCOE starts with IPsoft and our client’s developers and engineers. This small team works together to upload an assortment of documents designed to enrich Amelia with new knowledge. This includes a company’s business vocabulary, employee handbook, company directories, and historical customer support data. Clients provide IPsoft with their industry-specific semantic network, and the logic framework they’d like Amelia to follow in her decision-making.
Clients also provide IPsoft with pre-trained classifiers and analytic modules that help Amelia meet industry requirements and regulations. The combined team then auto-annotate Amelia’s conversational framework so she can understand utterances and common phrases. These initial processes provide Amelia with the data and structure she needs to become an expert in her role.
Once Amelia has been fed all of the information she needs to do her job, our clients work side-by-side with IPsoft’s Conversational Experience Designers to create the most human AI experience. If our client is an insurance agency, our client will need to dedicate a team of employees to come up with a range of dialogue possibilities, including tangents people might take during conversations, and idioms people might use when speaking with Amelia about insurance-related topics. The same rule applies to any industry in which Amelia is hired.
Our clients also work with our Cognitive Implementation Engineers to ensure that Amelia has the appropriate emotional response to things customers say or do. What triggers might anger customers? Is Amelia dealing with acceptances and rejections? Will she be handling fraud-related issues? Once clients have established a comprehensive list of triggers, they can work with IPsoft to provide appropriate emotional responses.
For example, should Amelia remain firm when telling a customer that he’s been rejected for a policy? Should she take a more understanding and empathetic approach and offer alternative plans? Should she escalate angered customers to human agents? This team goes through every possible scenario and chooses the best responses to help determine how Amelia should use her emotional intelligence. Amelia’s responses to the many emotional landmines that may come her way when interacting with customers play a big role in how satisfying an experience she ultimately provides – which of course can impact recurring revenue and new business.
Testing Amelia’s Skills
Once Amelia is ready for customer interactions, a team of employee super users will join the CCOE. This group of back-office employees engages with Amelia to optimize her responses. This team will be focused on correcting mistakes or altering what may be too-robotic responses Amelia provides during her extensive testing period. They’ll test Amelia to ensure that she’s answering questions correctly, with emotional intelligence, and following the appropriate business processes. This is the time for clients to tweak and refine Amelia’s in-conversation skills to ensure that she speaks like an insurance agent, or a credit card claims processor, or a medical professional. If Amelia is being deployed as a digital assistant for call center or support agents who interact directly with customers, the super users can verify that Amelia is providing accurate and helpful information to those agents.
If Amelia encounters a request that isn’t covered by her training and she can’t complete a task on her own, she should be escalating requests to human colleagues, and super users in the CCOE can ensure proper handoffs. When Amelia escalates a conversation, she stays connected with her co-worker to determine how the issue was resolved. With our client’s permission, she will then apply that knowledge to any similar conversations in the future. Even after Amelia has been deployed as a customer-facing solution, the CCOE should continually monitor what she’s learned from escalated calls to determine how her abilities can be enhanced and which should continue to require human intervention.
After initial success and refinement with a small user group, our clients should expand the CCOE to a larger group of employees. If Amelia passes this final test, she’s ready to start working directly with customers.
The Purpose of the CCOE
Although this is an intensive and comprehensive training process, Amelia can be ready to handle basic tasks in just weeks, and training for a single use case will only take three months, if not less. The quality of Amelia’s training is what allows her to ultimately deliver positive business outcomes, which is why proper training through a CCOE is such a critical success factor.
A cognitive solution designed, trained, and monitored by a CCOE will always outperform a plug-and-play solution. That’s because cognitive AI is best used in collaboration with humans. Human employees are essential to approve, fix, refine, or reject any of the machine learning that the solution wants to apply to future business processes. When combined, Amelia and a CCOE deliver the best possible outcomes for your business and its customers.