Companies that use AI copilots experience numerous benefits, including streamlined processes, improved customer experience, and cost savings. Large language models (LLMs) form the basis of many AI copilot applications. While LLMs have a remarkable capacity to generate humanlike text, that same capability is the reason businesses need to exercise caution in order to prevent misapplications of this otherwise transformative technology.
OpenAI’s GPT might be the most recognizable example of an LLM. The interface for the model’s third iteration, better known as Chat-GPT, was what went viral nearly immediately after its launch in late 2022. While its creators might have anticipated the toolkit would gain traction and a modest following, they certainly didn’t expect it to reach mainstream popularity so soon after its release, reaching nearly 100 million users in just two months.
We have already seen LLMs make a difference across sectors in business-to-consumer (B2c) workflows. Companies have automated aspects of customer service, leading to improved wait times. Additionally, LLMs are used to analyze customer data, allowing businesses to provide personalized customer recommendations and related services, which has the potential to impact sales positively.
Emerj Senior Editor Matthew DeMello recently sat down to talk with Carlos Quezada, Vice President of Customer Experience Strategy, Automation & Enablement at Hewlett Packard Enterprises, on the ‘AI in Business’ podcast about building customer journeys simultaneously from the business-to-business (B2b) and infrastructure perspectives.
The article below covers how Hewlett Packard Enterprises (HPE) uses an AI copilot for customer onboarding and how they plan to expand that copilot’s use cases to include interactive conversations with customers, which serves as a compelling use case for business leaders, demonstrating is an excellent example of how an automated onboarding process can be paired the power of large language models.
We will also cover how HPE uses data throughout the entire customer lifecycle, including how to predict customer conversions. The article will focus on two critical takeaways from DeMello’s conversation with Quezada:
- Balancing automation with humans in the loop in critical workflows: Creating a human-led customer success model backed by machine learning and automation.
- Driving digital transformation across the customer lifecycle: Strategies for integrating data throughout the entire lifecycle to improve customer experience through journey mapping, from pre-sale to post-sale experiences.
Listen to the full episode below:
Guest: Carlos Quezada, Vice President, Customer Experience Strategy, Automation & Enablement at Hewlett Packard Enterprises
Expertise: Customer experience transformation, AI, machine learning, big data analytics, automation, diversity and inclusion
Brief Recognition: Prior to his current role, Quezada was Head of Digital Services Strategy & Customer Success at Aruba, a Hewlett Packard Enterprise company.
Balancing Automation with Humans in the Loop in Critical Workflows
At the start of Quezada’s appearance on the podcast, he describes the parallel between B2b and B2c spaces. Quezada mentions the approach of evaluating how customer success was achieved in the B2c space and applying that to the B2b space.
Hewlett Packard looks at how to drive digital transformation and customer engagement when there is a three-tier distribution system. Subsequently, they translate lessons learned from the software business, or rather software and hardware in their case, and apply that knowledge to interactions with their distributors and, ultimately, the end customer.
Prior to his current role at HPE, Quezada used AI and machine learning to create innovative service models to drive customer experience and business outcomes at Aruba, a security and networking subsidiary of HPE.
He explains how determining customer success in Aruba is based on a layer model: land, adopt, expand, and renew. The focus has always been on the land phase and how to accelerate time to value and get the customer through onboarding and then start adopting; they’ve built out the mechanics of that process.
He mentions how, over the last two years, there has been a shifting left concept in his organization. Essentially, a lot of the capabilities they originally built for customer success to onboard a new customer are very relevant in the pre-sales phase, and he acknowledges how unfortunate it would be not to share that with the rest of the company.
As a result, his team has undergone a recent shift where they have brought the framework and data-focused thought process they use in Aruba into the rest of the business.
Quezada distinguishes technology and framework. He explains, “I think in situations where technology is maybe not available, there are processes in place to overcome that lack of technology kind of.”
He tries to understand what the current state is and what the friction points are. He then determines what data they have and what kind of assumptions can be made. As part of a longer-term strategy, he figures out ways to insert technology into the process to automate and accelerate.
When asked about whether or not AI will close off loopholes on repetitive tasks or provide multiple alternatives, Quezada explains that it comes down to a level of apprehensiveness that many people have. He doesn’t think anyone is really ready to give cognitive AI and ChatGPT free reign.
To provide context, he explains how his organization has taken its digital engagement map and personified it. As a first step, Quezada’s team created a fictional, animated character that assists with onboarding. Initially, this fictional character was confined to a video, but then it became more interactive, and now the character does month-end reviews.
Based on initial progress, they plan to test out a copilot version. He explains how they can identify use cases where the copilot can be unattended and have interactive conversations with customers on specific topics. He foresees a lot of copiloting examples in the coming years; he thinks eventually, the issue of trust will be overcome, and it will become more common.
Driving Digital Transformation Across the Customer Lifecycle
Quezada continues to explain how Hewlett Packard focuses on its journey mapping exercises, which would typically start with the onboarding phase after the customer purchases something. Thanks to their team’s agility in the lifecycle, they’ve looked at what they can do when the customer is in the pre-sales phase of learning and trying.
He explains how they recognize the benefit of information they collect about a customer’s adoption and consumption pattern after they’re a customer. He emphasizes how it is very relevant for someone who is going through a trial.
His team takes that model and makes that information available to inside sales reps. Their approach provides sales reps with visibility into how often a customer is logging into a demo system and what features the customer is using. As a result, they can start predicting whether or not the customer will convert.
Data is helpful throughout the entire lifecycle. Quezada compares his situation to the Amazon experience, emphasizing that even during procurement and shipping, it’s beneficial to know how many steps away the server is at all times. He continues to describe how it’s essential to leverage data that already exists to provide that level of transparency to improve the experience of customers.
In the process, Quezada notes that he is very aware of the influence B2c has on B2b. “I do think that B2b is learning a lot from B2c.” He further explains that he’s been stretching the capabilities of a lot of the vendors that he works with to push to bring some of the B2c experience to B2b.
He goes on to explain how the initial customer success program at Aruba was primarily a grass-roots effort. Quezada also doesn’t hesitate to mention how he was criticized from an industry perspective for his digital-first approach to customer success.
On the consumer goods side of logistics, historically, to achieve customer success, companies would hire multiple CSMs and manage accounts for a portfolio of customers. Because HP didn’t have resources or a large budget, Quezada leveraged his previous background in big data analytics and created a digital-first approach.
His team focused on how to address the extended tail portion of customers and realized that 96% of their customers didn’t warrant a human touch because of the dollar value of their account. However, when considered in aggregate, 96% of customers actually accounted for more than 50% of the revenue of the product.
They knew they had to pay attention to those customers but weren’t able to devote humans to them. To determine where automation fits in, he says it’s a segmentation exercise. Customers that are large enough still require a human presence, but smaller customers still should feel like they’re your only customers.
From an early point, they’ve honed in on a segmentation model and developed a three-tier engagement model:
- Digital only
- Hybrid
- High Touch supported by Digital
Quezada concludes by noting that his team has automation and AI supporting them in the infrastructure, but their customer-facing operations are very much still human-led.
