Miko uses AI to scale CX for high growth and assist busy agents

Miko is taking childhood learning to a whole new level. The advanced consumer robotics lab created a series of AI-powered robot companions that educate, engage, and entertain kids. The company’s latest robot, Miko Mini, features an emotive face and voice to create meaningful interactions, along with new voice-controlled activities and a range of educational STEM (Science, Technology, Engineering and Math) content. The company also sells automated chess boards.

Founded in 2015, Miko is now home to the world’s leading educators, engineers, psychologists, and content developers, with offices in Silicon Valley and Mumbai.

Meeting seasonal demand

Although Miko’s applications and programs are designed for kids, it is parents who contact customer support for assistance. If a product issue arises, parents may feel a sense of urgency to get their questions answered and get the product working as fast as possible.

Miko products are also a popular choice for holiday gifting, with a significant portion of the company’s sales made during the holiday season. This increase in sales generates a large wave of support tickets and adds pressure to deliver great customer service, because a negative support experience for parents could lead to a negative holiday experience for their children.

Growth brings new CX challenges

Creating positive support interactions is top of mind for Sandeep Jindal, Vice President of Customer Experience at Miko. He focuses on providing great customer support through self-service during the peak season, which is becoming more challenging as the startup grows.

“Every year, we see our growth projections increase four to five times,” says Jindal. “While we’ve made significant progress and are solving over 90 percent of queries via self-support, the other 10 percent of manual support volume will just keep multiplying year after year.”

As the customer experience team strives to keep up with a 12x to 15x spike in holiday volume, Jindal is also focused on resolving support issues efficiently.

“When I think about how to cover the gap in manual support and reduce handling time, training becomes super critical,” says Jindal. “Are agents properly trained to understand the process and the product? Are agents following the correct troubleshooting steps and replying to customers as soon as possible? Even if I can save 10 seconds per ticket, I want to do that.”

Scalable channels and smoother ticketing

Based on past experience using Zendesk, Jindal made a strong case to get Miko off Salesforce Service Cloud and replace Kapture for ticketing.

Implementing the Zendesk Suite for customer experience has unified workflows and improved visibility across departments. Support ticket handoffs used to happen via Google Chat or email, which made it difficult for other agents to track ticket progress. Now, the logistics and repairs teams use Zendesk to follow up on child tickets, and the product team can view support tickets and apply customer feedback in beta testing.

While voice was the most popular channel two years ago, it wasn’t scalable given the annual surge in support tickets. Using Zendesk enabled Miko to direct more customers to the help center and eliminate inbound phone support, though customers can still request a callback.

Around 45 percent of daily support volume is now handled through Zendesk chat, with 40 percent through email and web forms. Adding Zendesk AI, powered by Amazon Bedrock, allowed the company to expand its AI agent functionality and launch Miko Chatbot, ushering in a new era of intelligent support.

“Zendesk AI exceeded my expectations”

Jindal was motivated to find an AI-powered solution with conversational abilities when he saw what ChatGPT could do. After six months of evaluating vendors, he found that most AI tools were good at handling basic sales-related inquiries, but not complex customer service issues.

“Zendesk AI exceeded my expectations and the capabilities offered were much better than what I’d seen with other products on the market,” recalls Jindal, who got Miko to join the early access program. “Zendesk AI allows me to choose what I want to do when we detect a particular kind of a customer intent. Being able to customize that setup was really empowering.”

Transforming self-service

Miko products are delivered to more than 140 countries worldwide, making 24/7 support essential. The Miko Chatbot, powered by generative AI, offers a simpler way for customers to self-serve, especially during peak season.

“Customers should be able to see exactly how to interact with the bot and how to troubleshoot any kind of issue. We want to make it easy for customers and we want to keep the workload of our agents manageable,” says Jindal.

Since launching Zendesk AI, Miko’s self-service rate has jumped from 75 percent to over 90 percent. Although AI was a major contributor to the impressive increase in self-service, the company also continually worked on improving Help Center articles and created easier support processes which also helped in attaining over 90 percent self support rates.

Without these improvements, Jindal estimates that Miko would have received approximately 7,500 tickets daily during peak volume days during the most recent holiday season. Instead, peak volume reached only 1,250 tickets a day. That level of deflection reflects major CX efficiency and cost savings for the company.

“As a CX leader, it’s important for us to not only improve customer experience, but also to keep cutting costs. In order to balance these two deliverables, we need to use technology to ensure that customers are empowered to self-support as much as possible,” adds Jindal.

Intelligent triage and agent assist

Jindal’s team has optimized support workflows by leveraging intelligent triage and intent detection in AI agent and email interactions. Next, they plan to create intent-specific emails that provide an immediate response and align with customer needs, rather than sending an auto-response.

Generative AI is helping agents craft better replies and solve issues faster, too. “The agent assist feature is really important for support, because it gives agents the ability to quickly expand their response into a full-fledged email,” Jindal says.

Miko’s Program Manager of Customer Experience, Souparnika Hebbar, uses AI in her work designing the help center. When sharing articles with customers, Hebbar used to rely on ChatGPT or input from colleagues to help her compose messages.

Now, she’s a fan of the expand and tone shift features in Zendesk AI. Agents can quickly make their responses more professional or friendly. For example, a support conversation about a fun game or feature could use a friendly tone, whereas a troubleshooting discussion might be more formal.

A better agent experience with AI

Jindal expects that Zendesk AI will continue to benefit Miko’s customer service strategy, as artificial intelligence becomes more human-like and evolves.

From his perspective, the more intelligence that can be built into the system, the more value it adds. “Data analytics is another area where we want to see AI coming to life. If support managers could view indicators of ticket trends beyond the standardized dashboards, that would help us react even faster.”

Miko already enhances its CX program with the Zendesk WFM (formerly Tymeshift) workforce management solution and AI-powered quality management Zendesk QA (formerly Klaus). Next, the company is looking to add Zendesk’s Advanced Data Privacy and Protection (ADPP) features.

“I’m really happy with the way Zendesk is evolving and not only coming up with new product features, but also acquiring different companies which can help increase their capabilities,” says Jindal. “The Zendesk team is doing great work, which has given me all the more reasons to be happy with our decision to use the software.”