For decades, we have learned how to use computers. What happens when computers learn how to interact with us instead?
That question shaped my conversation with Beerud Sheth on Dilse Omni Talks. Beerud is the Co-Founder and CEO of Gupshup, a global conversational AI platform for marketing, commerce, and customer support. An IIT Bombay and MIT alumnus, he previously co-founded Elance, which later became Upwork. At Gupshup, he has led the journey from messaging to conversational AI.
I wanted to understand what that journey means for the relationship between consumers and technology. We have moved from desktops to the internet, from the internet to mobile, and now toward AI. Is this simply another interface upgrade, or something more fundamental?
Beerud’s answer turned our discussion into a masterclass: the next major shift is not about fitting more functionality onto a screen. It is about allowing people to express what they need naturally.
Table of Contents
Lesson 1: Every Technology Era Creates a New Front Door
Lesson 2: AI Could Become the Primary Customer Interface
Lesson 3: Voice Is About More Than Convenience
Lesson 4: We Are Moving from Keywords to Context
Lesson 5: Easier Interfaces Make Better Questions More Valuable
The Real Shift: Less Navigation, More Understanding
Lesson 1: Every Technology Era Creates a New Front Door
Beerud began by simplifying several decades of technology into four phases: desktop software, websites, mobile apps, and conversational AI.
In each phase, businesses built around the dominant way consumers accessed digital services. First, they needed desktop applications. Then they needed websites. Then mobile apps became essential.
But beneath those changes, the interaction model remained remarkably similar. A customer had to understand a layout, find the correct menu, select a button, and follow a predefined journey.
“They would force humans to behave like computers.”
Beerud Sheth, on traditional digital interfaces
That observation stayed with me. We tend to celebrate how much easier interfaces have become, but even a well-designed app asks the customer to learn the system’s logic.
A touchscreen made pointing and clicking more convenient. It did not fundamentally remove the need to navigate.
Different screens, but the same underlying requirement: customers must navigate a structured layout.
The Four-Era Interface Framework
Beerud’s explanation offers a simple framework for understanding the transition:
Desktop era: The customer learns to operate an application.
Web era: The customer navigates pages, links, and forms.
Mobile era: The customer taps through an app’s screens and journeys.
Conversational AI era: The customer describes a need through messaging or voice.
Desktop → Web → Mobile → Conversation
The first three phases primarily changed where the interface lived. The fourth could change who does the adapting.
Instead of asking the customer to understand the computer, we can increasingly ask the computer to understand the customer.
Lesson 2: AI Could Become the Primary Customer Interface
Beerud believes messaging and voice conversations will become the primary way businesses and consumers interact. In his view, conversational AI agents could become a brand’s main front door, ahead of its website, app, or even physical location.
That is a significant prediction. It moves conversational AI beyond a support widget placed in the corner of a website. It suggests that conversation itself could become the starting point for accessing a business.
For me, the important distinction is between adding a conversational feature and designing a conversational experience. One sits inside the existing journey. The other changes how the journey begins.
Beerud connected this to changing consumer expectations. People are becoming familiar with tools such as ChatGPT and Claude. They are learning that they can describe a problem and receive a response without first identifying the correct page or category.
Those expectations will not remain confined to general-purpose AI tools.
Search provides a useful example. Beerud pointed to the emergence of conversational AI experiences within search, where the expectation is shifting from a list of destinations toward a direct answer.
If the customer’s goal is an answer, the interface becomes less important than the interaction. A conversation can take place on a website, through messaging, or through voice. It does not depend on one particular screen layout.
This is why I see the discussion as an omnichannel question, not merely an app question. We explore that broader shift further in our discussion of how conversational commerce is changing customer engagement.
Does That Mean Apps Will Disappear?
I would not reduce Beerud’s argument to “apps are finished.” His bigger point is that the customer’s primary entry point could change.
A business may still have a website and an app. But customers may increasingly expect to explain what they want rather than work out where that functionality sits.
The strategic question is not only whether your business has an app. It is whether customers must learn your app before they can get help.
Lesson 3: Voice Is About More Than Convenience
At this point, I brought the discussion closer to home.
There are three generations in my household: my parents, my wife and me, and my child. The people doing most of the typing are my wife and me. My mother uses voice input, converts it into text, and sends it. My son prefers voice notes.
That everyday observation makes the interface shift tangible. Different generations may arrive at voice from different starting points, but neither necessarily sees typing as the default.
Beerud took that example further. Keyboards, mice, screens, and touchscreens were all tools for controlling computers. Using them required some level of familiarity with technology.
Conversation changes that starting requirement.
“Computers can meet humans where they are.”
Beerud Sheth, on the promise of conversational AI
People learn to communicate long before they learn to operate software. If AI can respond in a person’s language and dialect, it could lower not just the digital literacy barrier, but also the reading and writing barrier.
Beerud described the possibility of reaching an elderly person in a remote part of India or Brazil through an interaction that feels familiar: speaking.
I took this as a powerful vision for digital inclusion, rather than a claim that every access challenge is already solved. His central point was about the interface barrier. Digital services become more approachable when people can engage on their own terms.
A Simple Inclusion Lens
I would summarise this part of the masterclass through three questions:
Language: Can the person communicate in a language or dialect they naturally use?
Literacy: Does the interaction require reading and writing, or can it happen through speech?
Digital familiarity: Must the person understand menus and navigation before expressing a need?
This is my synthesis of Beerud’s reasoning, not a named framework he presented. It helps make his argument practical: design around the person’s ability to communicate, not only their ability to operate a screen.
Lesson 4: We Are Moving from Keywords to Context
The shift is not only about speaking instead of typing. It also changes what we ask.
I shared a gifting example. Earlier, I might have searched for a suit to give my sister for Rakhi. The category was already decided, so the search mainly helped me locate options.
In a conversational interaction, I might start by explaining that my sister is 19 and that I need help choosing a suitable Rakhi gift.
Now I am not simply looking for an item. I am asking for help interpreting a situation.
Category-first interaction: Find a product I have already chosen.
Context-first interaction: Understand the person and occasion, then help me choose.
That distinction matters for commerce. Customers do not always arrive with a neatly defined product requirement. Sometimes they arrive with an occasion, uncertainty, or an incomplete idea.
The opportunity is to let them express that context without forcing it immediately into a category. This connects naturally with the broader evolution explored in Decoding Conversational Commerce.
Lesson 5: Easier Interfaces Make Better Questions More Valuable
Beerud then introduced an important qualification: being able to talk to AI does not automatically make everyone equally effective at using it.
Human communication works the same way. Some people explain themselves clearly, ask sharper questions, and guide a discussion more effectively. Those differences do not disappear when the other participant is an AI system.
“Prompting is a skill.”
Beerud Sheth
As AI takes on more coding, mathematical, and mechanical tasks, Beerud believes the ability to ask the right question in the right way becomes increasingly important.
That prompted a useful distinction for me: access and effectiveness are different things. Conversation can make technology accessible to more people. Clear communication can help those people get more value from it.
My Practical Takeaway: Need → Context → Guidance
Building on our gifting example and Beerud’s emphasis on prompting, I would use this simple communication framework:
State the need: Explain what you are trying to achieve.
Add the context: Share the details that influence a useful answer.
Guide the conversation: Clarify or redirect when the response does not address the real question.
This is not about memorising complicated prompts. It is about becoming more deliberate in how we communicate intent.
Beerud’s closing thought brought the discussion together: artificial intelligence and natural intelligence can be more effective in combination than either alone. Human direction still matters.
The Real Shift: Less Navigation, More Understanding
I began with a question about how interfaces have evolved. I came away thinking about how much effort we have historically asked customers to make.
Beerud’s masterclass was a reminder that the next interface may not be another destination to learn. It may be a conversation that begins with what the customer needs.
For business leaders, that creates two responsibilities: make the interaction more natural, and remain clear about the human context it must serve.
The opportunity is not simply to build an AI interface. It is to stop making every customer think like a computer.
I am Saurabh Agrawal and we come with a new episode on Dilse omni talks every fortnight and cover different aspect of omnichannel with amazing speakers.
