Delight.ai Launches 2 AI Agents

Delight.ai, Sendbird's customer agent subsidiary, today introduced two artificial intelligence agents: Delight Customer Agent, to handle shopping, booking, support, and retention, and Midas, a self-service AI agent management tool that lets teams build, test, and continuously improve their Delight Customer Agent using conversations.

Delight Customer Agent sits across the entire customer lifecycle so businesses can maintain customer context across all interactions and channels to help customers discover products, complete purchases, resolve support issues, and return for their next purchases without forcing them to start over. It understands customers' identities, preferences, and histories, reaches customers across channels, and can take actions while operating within company boundaries, approvals, and oversight.

With Delight Customer Agent, businesses can do the following:

• Guide product discovery and purchases;
• Qualify leads and book meetings, appointments, and jobs;
• Resolve issues and questions; and,
• Renew, re-engage, and grow customer relationships and lifetime value.

"Brands are at risk of recreating the same organizational silos with AI that have frustrated customers for decades," said John Kim, co-founder and CEO of Delight.ai, in a statement. "Customers do not care which department owns shopping, support or retention. They have one relationship with a business and expect AI to understand their relationship."

Midas, meanwhile, is an AI agent that enables any team within any department to build and continuously improve their Delight Customer Agents. Teams can describe what they want the Delight Customer Agent to do in natural language, and Midas will translate that request into an actionable process for the agent to follow.

Midas works across the following four stages:

• Build: turns workflows and policies into a process the Delight Customer Agent can follow;
• Test: runs proposed changes against simulated customers so teams can see how the agent responds before a real customer does;
• Improve: implements approved updates, including new policies, offers and requests; and,
• Analyze: reviews conversations and results to identify where the Delight Customer Agent falls short and recommends what teams should improve next.

"Every team has a running list of things they want their AI agent to do better, but today, every change becomes an engineering problem," said Nick Singh, head of product at Delight.ai, in a statement. "Midas changes that, and instead of waiting days for an update, those closest to the customer can make necessary changes in hours and can keep making updates so the same agent becomes smarter and more capable over time."