How Summarization AI Eliminates After-Call Work

Let’s face it…contact center leaders are maniacal about reducing average handle time (AHT) in service-oriented contact centers. Agents spend significant time manually capturing and summarizing customer interactions during their after-call work (ACW) time. The average wrap-up time at the conclusion of an interaction is six minutes, which means 6 more minutes on average is added to the handle time¹. What do agents do during those 6 minutes? They document call details, update systems, and plan follow-ups. All of that is important, but that’s additional time and money spent in the contact center, leading to a reduction in agent efficiency and less calls being handled per agent. This additional work can also contribute to high agent burnout². Bottom line, excessive ACW increases average handle time (AHT), directly impacting operational costs. So how do you solve this problem? Enter Summarization AI.

Launching Summarization AI Enhancements: Customization and Automation Tailored to Your Business

I am pleased to announce the next generation of our Summarization AI product.

With this launch, contact center leaders now have complete control to customize and fine-tune their summaries and after-call work notes using natural language prompts - all powered by Gen AI within a self-serve builder interface.

Observe.AI Summarization unlocks numerous opportunities, including:

Generating custom summaries you need from calls by simply using natural language prompts (e.g., answers in paragraph or bullet points)

Extracting specific information and storing those values (entities such as competitor names, offers or products mentioned during the interaction)

Classifying calls (pick from a set of values)

We also have Pre-built Summary models out of the box, including full and short summaries.

Eliminate After-call Work Completely

A big part of our ethos is to work with our customers to make sure they get value out of our products. Gen AI-powered products are no different. “We are proud of our human-powered approach with trust and empathy at the core,” said Ardie Sameti, Sr. Director of AI and Automation at Accolade. “By customizing and co-developing with the Observe.AI platform, we can analyze 100% of our interactions with members and provide real-time feedback to our Care Teams with areas for improvement. We are able to create high-quality after call summaries using generative AI that are consistent and actionable. Our work with Observe.AI helps our Care Teams optimize their time and provide better member experiences.”

I personally love this example because it highlights the end to end nature of Summarization AI capabilities, from generating a highly complete and correct summary to directly pushing it into a CRM via Observe.AI Integrations, thus fully eliminating ACW. That’s how you get your 6 minutes back.

Everyone Does Summarization…So What’s the Big Deal?

Leveraging Gen AI for summaries - powered by a large language model (LLM) - is an incredible use case for contact centers to reduce or eliminate after-call work. However, you can’t just throw any LLM at the problem. Research shows that summarization using generic LLMs does not perform optimally. Think about all the nuances of a contact center: errors with transcription, disfluencies (um, ah, etc), background noise, agent vs. customer dialog. Generic models are great at creating awesome sounding content and adding variation. But let me ask you, do you think that your customers want you to vary up your call notes just to make sure the summary sounds awesome? Or do you think they want the action items to be complete and correct so that next time they call in they don’t have to repeat themselves? My guess is the latter. That is why our Observe.AI contact center LLM, trained on 2 billion interactions representing 100+ years of contact center-specific tasks and data, powers our Summarization AI. Our LLM indexes on correctness and completeness, leading to a 35% higher summarization accuracy than generic LLMs.