Context engineering and the Future of AI-powered business – Crypto News

Getting customers in the door has never been easy. But as businesses across industries are discovering, the real challenge begins after that first transaction. Whether in B2B or B2C markets, retaining customers has become significantly harder than acquiring them, as expectations for speed, personalisation and seamless service continue to rise.

Artificial intelligence is increasingly being positioned as the tool that can help businesses meet those expectations. Not just by automating responses, but by enabling richer, more human-like conversations at scale, conversations that help companies differentiate themselves in competitive markets.

That shift was the focus of Episode 14 of Mint’s All About AI series, where Abhishek Singh, Deputy Editor, Mint, spoke with Aditya Singh, Vice President, Product Management, Agent for Service, Salesforce; Prasad Raje, Senior Vice President, Product Management, Agent for Service, Salesforce; and Deepu Chacko, Vice President, Solution Engineering, Salesforce India. The discussion explored why businesses are now moving from prompt engineering to context engineering, and why this transition is becoming central to customer loyalty and long-term growth.

Watch the full episode below,

Why Retention Has Become Harder Than Acquisition

The conversation began with a simple but increasingly common observation: today’s customers are impatient, highly informed and quick to disengage when experiences fall short. Retaining them requires far more than efficient transaction handling.

Aditya Singh explained that customer expectations have fundamentally shifted. “In today’s age, customers’ expectations are increasing. With AI, you can really take advantage of AI and allow service agents and service leaders to stay on top of things and deliver that personalised experience so that it helps them build that trust with their respective brands,” he said.

Personalisation, the panel agreed, is no longer optional. It has become the baseline against which customers judge service quality, whether they are dealing with an airline, a bank, a telecom provider or an enterprise software company.

The Customer Journey Is Continuous, Not Transactional

Prasad Raje reframed the idea of customer acquisition itself, arguing that businesses often think about it too narrowly. “You are acquiring customers all the time. It’s not a one-time thing, because your goal as a provider or as a brand is to make that customer a lifetime customer,” he said.

Customers, he explained, are constantly cycling through moments of buying, being supported and buying again. Service and sales are therefore not separate journeys, but overlapping ones. Many service interactions, from order changes to support calls, function as sales interactions in practice.

From Salesforce’s perspective, this is why customer engagement must be approached as a unified strategy across sales, service and marketing rather than as isolated systems. Fragmentation breaks continuity, and continuity is what customers increasingly expect.

How AI Fits Into This Evolution

To understand AI’s role, Raje placed it within a longer history of technology change. Customer service once required physical presence, visiting an office to place an order or pay a bill. The internet and web transformed that, enabling remote transactions. Cloud platforms standardised and scaled those interactions, while mobile made them ubiquitous.

AI, the panel argued, represents the next major shift. “This is the first time computers can speak back to us in a language that is truly human,” Raje said, pointing to AI’s ability to engage in natural conversation, something earlier systems could not do.

That conversational ability is what makes AI so powerful in customer service contexts. However, language fluency alone does not create a good experience.

Why Prompt Engineering Is No Longer Enough

A recurring theme in the discussion was the limitation of prompt-based AI systems. While they may respond fluently, they often lack awareness of the customer’s situation.

Deepu Chacko emphasised that scale is the real constraint. “Every consumer is expecting a personalised service, and to do that at scale is something that can only be done with AI,” he said.

But personalisation at scale requires far more than good prompts. It requires systems that understand business rules, brand tone, policies and customer history, and that can apply them consistently.

Raje illustrated this with a simple airline example. “When you’re calling Air India, you expect the agent on the other side to have context about the ticket you just booked, or the flight you have taken, or the baggage issue you already reported,” he said.

A generic AI may speak well, but without access to such information, it cannot respond meaningfully. “That raw ability to sound human has to be put together with the context of the business,” he added.

Context Goes Beyond Data, It Includes Emotion

Context, the panel stressed, is not limited to transactional data. It also includes what is happening in the moment.

A customer may begin a conversation calm and become frustrated or anxious as it progresses. Systems need to recognise that emotional shift and respond accordingly, whether through reassurance, an offer or escalation to a human agent.

This is where the hand-off between AI and humans becomes critical. Human agents must be equipped not just with the customer’s historical data, but with the full conversational trail that led up to the escalation.

AI Already Operating at Scale

Abhishek Singh cited Air India as a real-world example of how far AI-led service has progressed, noting that the airline’s AI agents now resolve the vast majority of customer queries, with only a small fraction requiring escalation to human agents.

The example underscored how AI has moved beyond experimentation. “We are well past the point of showing proof of concepts and demonstrations,” Raje said, pointing out that such deployments are already working across industries and use cases.

Addressing the Fear of Hallucination

One of the most common concerns businesses raise about AI is hallucination, the risk of incorrect or fabricated responses.

Aditya Singh acknowledged this hesitation directly. “Humans give wrong responses as well. It’s not just AI that hallucinates,” he said.

Raje expanded on this by explaining how expectations around technology have been shaped by decades of deterministic systems. “AI is not deterministic. It is probabilistic and closer to humans,” he said.

The key to managing this risk, he argued, lies in context. “If you give the AI agent the right context, it has much less propensity to hallucinate. Output can only be as good as the input,” he said.

That context includes accurate data, proper workflow integration and clearly defined organisational guardrails.

What Businesses Are Asking for Next

Looking ahead, Raje shared what customers are increasingly demanding. “Customers are asking if we can convert six-minute conversations into two-minute conversations, or make their call centres operate 24 hours instead of 9 to 5,” he said.

Businesses want richer engagement without the prohibitive costs of scaling human-only teams. AI, the panel agreed, makes that expansion economically feasible.

AI’s impact is also spreading beyond customer service. Across sales, marketing, operations and billing, repetitive tasks are being automated so employees can focus on higher-value work.

Chacko added that AI is also democratising capability, reducing the need for deep technical expertise, enabling faster analysis and supporting real-time language translation, an especially significant development in a diverse country like India.

A Lighter Look at What’s Coming

As the session wrapped up, the conversation turned lighter, offering glimpses into how AI is already reshaping everyday experiences.

Aditya Singh pointed to self-driving taxis in San Francisco as a striking example of how far AI has progressed in the physical world. Raje, meanwhile, spoke about rapid advances in AI-driven video generation, highlighting how these systems now demonstrate an implicit understanding of real-world physics.

Chacko shared a personal example of using AI to plan a complex international trip, aligning schedules, routes and travel preferences in minutes, a task that would otherwise have taken far longer.

When asked what they would automate in their own lives, the answers ranged from managing Slack messages to grocery shopping and even meal planning, a reminder that the ultimate ambition of AI is not just efficiency, but anticipation.

Context as the Foundation of the Future

As the discussion concluded, one idea stood above all others. Prompt-based AI can answer questions. Context-aware AI builds relationships.

By combining historical data, live interaction signals and emotional awareness, context engineering allows businesses to move beyond transactional service towards trust-based engagement. In a world where loyalty is fragile and expectations are unforgiving, that shift is no longer optional.

It is fast becoming the foundation of how modern businesses operate, and how they retain the customers they work so hard to acquire.