AWS has released an AI-enhanced email workflows feature to automate customer service.
This feature is designed to further the Amazon Connect Email platform, utilizing built-in capabilities that enable agents to deliver quicker response times.
The workflow tool is the latest addition to AWS’s email and contact center capabilities.
Released in November 2024, Amazon Connect Email is an omnichannel support feature that enables customer service agents to respond to and divert customer emails all within the same system as voice and chat, allowing agents to handle customer queries in a single space.
The AI-enhanced email workflow tool enhances Amazon Connect Email by allowing service agents to expedite email customer service through automation capabilities.
By utilizing large language models (LLMs), these AI-powered workflows can allow Amazon Connect to analyze emails, detect customer intentions, assess potential risks or complexity from the interaction, and evaluate next steps.
After analyzing the email, the tool will provide a summary of the customer’s profile and any previous activity with the enterprise, the determined query category, and a brief rundown of the email to help tailor the response accurately.
This tool also grades the received email based on how confidently it understands the message with Amazon Bedrock API and Claude AI, factoring in clarity, tone, topics, risk assessment, and time sensitivity, while also considering whether the customer is part of any premium packages by retrieving the user’s profile.
This profile retrieval will include the customer’s current credit score, service level, and contact history to help the tool further evaluate the email score.
After this, the tool implements a two-step process, where the LLM produces binary outputs for each negative factor it recognizes, using an embedded mathematical function to ensure the calculations align with the scoring evaluation.
Enterprises can also personalize the scoring framework to fit their needs and determine the number of emails routed to an agent.
Once a score has been determined, the email will receive either a generated response (if the score is 80 or higher) or be routed to an agent’s inbox for a personalized response (if the score is lower than 80), whilst also providing an explanation detailing how the tool determined the score.
To simplify future interactions, the tool can automatically generate a case detailing the email and information previously gathered, allowing agents to view conversation history from a single place where email handling has occurred with generative AI.
Agents can also personalize AI-generated responses to keep human intelligence in the loop and can add AI-powered workflows to Amazon Connect Email via Amazon Bedrock.
What This Tool Means For Customer Experience
The Amazon Connect AI-enhanced email workflows allow enterprises to bridge the gap between customer demand and agent availability through automating repetitive email tasks and filtering complex queries to agents, providing customers with human responses and agents with enhanced productivity where needed.
These AI-powered workflows outrun traditional automation systems by understanding context clues behind interactions, including emotion and characteristic human responses by analyzing a customer’s profile.
This avoids agents from partaking in repetitive research tasks and instead solving complex, human-based problems to deliver solutions that an LLM cannot solve, resulting in higher levels of meaningful work for the agent.
This also resolves issues with email backlogs, customer survey responses, and agent exhaustion, driving improvements in customer service.
