Rethinking AI for Sales, Training & Customer Service

Understanding Sales and AI Agents

I first called this newsletter Searchlight Sales. I based this on Karl Popper’s insight that we seek knowledge rather than it being given to us. We absorb information by forming concepts of what we are reading, hearing or watching. My idea was this is how customers arrive at buying decisions.

As a consequence, the sales process is a guided discovery. People buy once they form the concept of a solution to their problem. The person who helps them create that concept gets the sale. AI support tools should adopt this approach.

There have been challenges with chatbots flattering users of late. They work as a servant fetching information, who bows and scrapes at the master’s instructions. The problem comes when students, coders and writers, copy and paste what is given them. Not only is information often incorrect, but users miss out on the process of creating knowledge. Rather than encourage critical engagement, many AI tools today reward surface-level mimicry.

There is no need for AI to work this way. The Socratic method is a means of instruction that seeks understanding through deep questioning. More than 2,000 years after it was devised, my tutorials at Oxford used a similar technique to help students embed key concepts. AI tools should be designed using this same principle of guided discovery to foster durable understanding.

Reimagining Onboarding Through Dialogue

One of the biggest chores managers face is onboarding new colleagues. It shouldn’t be this way, because getting the best out of people is the purpose of management. Yet repeating the old routine is tedious and often outsourced to junior staff, who have less understanding of why things work in a particular way.

Automated onboarding takes the form of a series of videos. Each video ends with a short quiz and the software tracks completion. Most employees forget the content soon after finishing.

Imagine interactive onboarding that showed colleagues a process and then asked them why it works that way. This is a guided conversation rather than a multiple choice quiz. We are not testing whether people have watched a video. We are imparting an understanding of why the company works a particular way.

This week I am taking the latest Agentic AI training from NVIDIA. The company supplies the building blocks for teaching agents that could guide students through a curriculum. Yet even it still uses videos and quizzes to test my understanding.

I would prefer an agent through which I could question the materials, ask for clarifications and receive succinct answers, before resuming the course. In turn, the agent would ask me to explain concepts to ensure I had understood them. The two-way questioning works at my pace and level of understanding. Right now, I must jump between websites and training videos to catch up on elements that I do not understand.

When we understand a process we adopt it faster and stick closer to it. We make suggestions for improvements. New joiners might explore with an onboarding agent why past approaches may not fit their new environment.

Creating training materials for agents forces the business to reflect on its own practices. Senior staff will need to justify operations directly. When multiple trainees ask the same question, it often reveals flaws in the process. As a result, the quality of the onboarding improves.

Building Trust Through Enquiry

Tidio research shows that both businesses and customers want speed and availability from chatbots. Thereafter, purposes divide. Customers want the fastest path to information and associate that with human availability. Businesses want to automate repetitive questions and find opportunities to upsell new services.

The reason people prefer human help to check order status, or discover deals, is lack of trust. They know that a machine pushes them to what is best for the company and hope that a human will be sympathetic to their needs. It doesn’t have to work like this.

A customer service agent will answer questions and engage in conversation. It will ask why customers use a product, what others they considered and what part of the experience they would change. It can guide people to more suitable services.

I had to replace my WiFi last month. I called EE and answered a list of annoying questions about whether the box was plugged in. Finally, the human agent checked my records and saw I qualified for a faster and cheaper WiFi, which I was happy to accept. EE could use its website agent to handle this faster and at much lower cost.

Customers trust companies that reward them once in a while. I was pleased to call and find that I could have a better service for less money. Wouldn’t it build more trust if the website agent offered this without me having to ask?

In designing AI tools with real business value, we must shift from systems that merely dispense answers to ones that cultivate understanding. Whether training a new employee or guiding a customer, the aim should be to foster insight. This requires creating agents that listen, question and guide users to form their own conclusions. Only then will AI earn trust and deliver the kind of engagement that transforms user experiences.

Questions to Ask and Answer

1. Where in my business would a dialogue increase trust and understanding?

2. How can my onboarding and support processes teach rather than tell?

3. Do my automations reinforce critical thinking, or just save time?

Here are 3 ways I can help:

1. Explore AI use cases.

2. Book a consultation with an AI expert.

3. Hit reply to ask about how to build AI agents.