Making Agent Assist actually work for your Agents!

While Agent Assist technologies are not new, they have made a big splash with the advent of AI driven capabilities.  What started as agent scripting technologies that guided agents through a scripted interaction flow has now evolved to be dynamic, contextual and intent driven.  With AI driven support automation managing repeat tasks and queries, Agents and Supervisors are now required to handle more complex tasks. It is time to retire Agent Scripting - customers want to interact with agents only when all other means like self-service or automated bots fail.  When they do that, they require contextual help, and the last thing they need is an agent running them through a pre-canned script.  Simply annoying and ineffective!

So what makes Agent Assist useful? IriAgent has deployed AI Agent Assist capabilities for a large number of customers and here are a few lessons we share!

• Feed Agent Assist with your brand and compliance guidelines  - Brand and compliance guidelines reflect an organization’s values.  Agents need to follow these guidelines to have brand-compliant interactions with customers, but agents can sometimes forget to follow the guidelines. Make sure to feed brand and compliance guidelines to Agent Assist technology that prompts agents appropriately, and scans agent responses in real time and flags them if they are non-compliant with brand guidelines.

• Incorporate both, real-time sentiment and intent detection - It is not one or the other!  Both sentiment and intent detection are important in equal measure.  Most technologies may do sentiment detection, but what use is it to know that the customer is angry/upset (or happy) but not having means to address the situation.  Agent Assist technology needs to deliver both to truly help agents be their best.

• Continuosly feed Agent Assist with relevant cross-functional data - This is the most critical aspect of making agent assist effective - especially in the hi-tech vertical.  When an Agent interacts with a customer for a complex task -  they need help synthesizing information from different sources fast and in real-time to be effective. Following accurate intent detection, Agent assist capabilities need to quickly prompt the agent with,

  • Similar issues - i.e. historical information from CRM systems.

  • Relevant articles recently used for similar issues - i.e. relevant knowledge base articles.

  • Any immediate known engineering issues - i.e. understanding open JIRA items, Confluence articles or product enhancements within the engineering team.

  • Any known infrastructure issues - i.e. understanding application and infrastructure monitoring feeds from Devops in real-time.

  • Logs specific to the customer - i.e. information specific to the customer’s environment. This is applicable more for the hi-tech vertical.

Agents very much appreciate all of the above data points as this helps them quickly assess the situation and work on resolution or next steps rather than information gathering.

• Continuously measure and improve Agent Assist capabilities - It is important to recognize that Agent Assist needs to improve with time and keep incorporating relevant data for continuous improvement as mentioned above. Equally important to measure KPIs. We recommend establishing 3 to 4 of the following KPI’s as performance indicators,

  • Customer satisfaction score (CSAT)

  • First Contact resolution (FCR) 

  • Customer retention rate (CRR)

  • Net promoter score (NPS)

  • Customer churn

  • Number of upsells and cross-sells 

  • Customer lifetime value (CLV)

Customers today  expect effective, authentic and personalized experiences from Agents and agent assist technologies need to help agents be their best! We would welcome the opportunity to showcase how we can incorporate the above best practices with Agent Assist technology from IrisAgent AI Support Automation platform - sign up today for free!