Readintersections

Forty years in banking technology taught me that intelligence is only as good as the data it can trust.

Over the past year, I have met with financial institutions across multiple markets, and one conversation keeps returning. It is no longer primarily about AI budgets or which model to deploy. It is about the data underneath all of it.

The question in many boardrooms has shifted from, “How do we deploy AI?” to something more fundamental: “How do we redesign our enterprise data architecture so AI and autonomous agents can operate with trust, speed, and scale?”

That may sound like an incremental change. I do not believe it is.

In forty years of banking technology, it is one of the most consequential shifts I have seen. We are moving from systems that help people understand what is happening to systems that can interpret context, support decisions, and increasingly take action.

That changes the role of data architecture. It is no longer simply the foundation for reporting and analytics. It becomes part of the institution’s decision and execution infrastructure.

How We Got Here

I have watched this industry reinvent itself four times. Mobile-First brought banking to our fingertips. Digital-First reimagined customer journeys and operating models. AI-First enabled intelligence through personalization, automation, and predictive decision-making. Each wave solved a specific business problem and created significant value.

Now we are entering the era of Agentic Banking, where autonomous AI agents can reason, collaborate, make decisions, and carry out complex workflows across the banking value chain.  We are moving from systems that produce insights to systems that can act. If you are leading a bank amid this transition, that distinction changes everything about how you build.

The Problem Nobody Solved Along the Way

Here is what I have seen across almost every institution I have worked with. The pace of innovation has consistently outstripped the evolution of the underlying enterprise data architecture.

Data Warehouses gave way to Data Lakes, then Middleware, APIs, RPA, Hyperautomation, AI, and now Agentic AI. Each wave added capability. Yet, in many institutions, the underlying foundation did not evolve at the same pace.

 Many institutions today find themselves with segmented data estates, duplicated customer information, inconsistent governance, and AI models that cannot consistently access trusted, real-time enterprise data. You can build the most sophisticated agent in the world. If it cannot access a trusted and current view of the customer, it may make confident decisions based on incomplete information.  In high-stakes banking workflows, that can be more dangerous than keeping the decision under human control.

This is driving significant investment in cloud-native data platforms, real-time event-driven architectures, enterprise data products and data mesh principles, API-first integration, strong data governance and lineage, and unified customer and enterprise data models. In my view, this is becoming the most important technology investment of the decade.

From Systems of Record to Systems of Engagement

Another major shift is taking place in customer relationship management.

For nearly three decades, CRM platforms have served as the primary interface for sales, service, and customer management. In the Agentic AI era, that role is beginning to evolve. CRM is increasingly evolving into the system of record, while intelligent agents begin to take on a larger role as the system of engagement.

Instead of relationship managers manually identifying cross-sell opportunities, next-best actions, portfolio gaps, or customer service interventions, AI agents can continuously analyse customer behaviour, life events, risk, profitability, and intent, then recommend or orchestrate personalised actions in real time.

Customer Value Management is undergoing a similar transition. It is moving away from periodic, campaign-based analytics toward continuous optimisation, where intelligent agents can help balance customer experience, revenue growth, risk, pricing, and profitability.

But this shift depends on the quality of the underlying data. An agent cannot create a coherent customer experience if customer information is fragmented, definitions are inconsistent, or important signals are delayed. It also raises the bar for governance. Institutions will need clear accountability, decision boundaries, explainability, and appropriate human oversight as agents take on more consequential customer interactions.

The institutions that lead in the Agentic era will not necessarily be those with the largest AI budgets. They will be those that have built a trusted, cloud-native enterprise data foundation capable of supporting intelligent agents across the organisation.

In this environment, data architecture is no longer back-office infrastructure. It is becoming part of the enterprise’s decision and execution system.

The next competitive advantage will not come from AI alone. It will come from AI built on data that is trusted, connected, current, and governed.

Judgment is the New Infrastructure

I was going through the articles we have published on Read Intersections and noticed something. Six leaders, from six different corners of the working world, had arrived at the same conclusion.

Lindsay Jessup termed it Chaos Cartographers. Her argument was that technology produces beautiful, confident maps. What it cannot do is tell you which map is worth following. That call is still ours to make. And most organizations have not built the muscle for it.

Sandeep Naug used a different image. Everyone has a paintbrush. Not everyone is Picasso. Access to new tools is no longer what separates people. What separates them now is the craft knowledge they bring to the prompt. Without that, you get more work. You do not get better work.

Ashwini Suman said the same thing from inside an IT services company. His developers have not written a line of code in months. Their role did not shrink, but moved upstream to knowing what to build, which trade-offs are real, and what "done" actually means for the person paying the bill.

This issue wasn’t limited to IT. In manufacturing, Rajendra Abhange saw it on his factory floor. The sensors caught everything. The plant still broke down because there was no standardized human response at 2am when the right engineer was off shift.  The gap was judgment applied consistently.

Viiveck Verma watched as a global capability center tried to recreate its success in a new place. The manuals and dashboards arrived easily, but what truly mattered- the trust, the hard-won lessons, the quiet confidence between teammates- couldn’t make the journey on paper. He realized that while processes can be copied, true judgment is built face-to-face, over time, in the daily give-and-take of real relationships.

Joshua Konowe brought insights from the sports arena. He observed that machines can recognize patterns, but humans interpret context. The conversation happening in football, and everywhere else, is not really about the technology. It is about who still gets to call the shot.

Taken together, these accounts show that, beneath industry jargon and metaphors, the underlying challenge is the same. Judgment has become the scarce resource. Everything else is closer to a commodity than it has ever been.

It implies something about how organizations are built. Most were designed to scale processes and repeatable outputs. That worked when execution was the bottleneck, which is not the case now. Judgment cannot be bought, downloaded, or handed over with a playbook. It compounds slowly, through decisions people own and are held accountable for.

The question this leaves for anyone leading through the next three years is not whether to adopt AI. That question is settled. The question is where in your organisation judgment is actually being developed, who is exercising it, and whether you are treating it as the infrastructure it has quietly become.

Most organizations are still measuring the wrong thing.

Joy, Laughter, and Love: UX Design for the Next Generation of AI Products

Walk into any children's aisle, and you're looking at the toy industry's version of a cereal box: bright colors, a licensed character on the front, and a promise of fun that rarely survives the car ride home. The industry built itself on the same formula as junk food. It was engineered to be craved in the moment rather than something that nourishes. Parents are worn out from buying overpriced, low-quality, character-branded plastic junk. We have all had enough.

As physical AI moves from apps into talking machines in the home itself, we are about to find out whether we repeat that same recipe at a far more intimate scale.

I've spent almost thirty years building products, from broadcast animation to immersive experiences, games, and robotics. The last sixteen of them have been in AI. What I've watched happen, generation after generation, is abstract metrics stripping the soul out of craftsmanship and storytelling. Nobody sets out to build something that manipulates a child, but if your revenue depends on session length, return frequency, and time on device, those become your product's real goals, whether you name them that way or not.

​
Boardroom Decisions Should Not Solely Drive AI in the Home

The metrics that built broadcast and social media include viewership, watch time, daily active use, and return rate. All of these were designed for one purpose: maximizing attention to sell around it. That playbook migrated cleanly from television to apps to social platforms, because the business model stayed the same throughout. What's shifting now is that this exact scorecard is migrating into AI products that sit inside the home itself: companions, assistants, wearables, and generative tools built for children and families.

What happens when a device already knows how you feel, doesn't wait for a prompt, and can actually do something to help? That's the new interface. A television could be turned off. A phone could be put down. An AI companion is increasingly ambient. It is present in the bedroom, the kitchen, the car, always aware of our surroundings, our moods, our routines, remembering who we are. A genuinely new interface deserves a genuinely new set of KPIs, not the same attention metrics wearing a smarter costume.

In Sweden, advertising directed at children under twelve has been restricted for more than three decades, forcing toy makers there to answer a different question: if you cannot chase a child's attention to sell them something, what do you optimize for instead? New regulations limiting social media and AI access for children are now emerging in multiple countries, including the US. The direction is the same everywhere: less tolerance for products engineered to hold a child's attention, more scrutiny of what that attention is actually being fed.

​

What If the Device Measured a Life Well Lived?

Almost every metric the industry provides to investors, such as engagement, retention, and session length, is a proxy for how much of a person's time and attention a product can capture. None of it asks whether a child or a family is better off for the time spent. Those are not the same question, and the gap between them is where family tech keeps losing its way.

What would you actually pay for a product that measurably improved your child's life prospects? It's part of why Harvard, MIT, and Princeton are so sought after: attending them lowers the odds a child struggles later in life, and parents already pay enormous sums for that outcome. We should expect the same from the technology sitting in our homes.

Social media is the warning tale. It started from a genuinely good place and can still hit every engagement target on its dashboard while leaving a child lonelier, a parent more distracted, or a family less connected than before. Nobody built that outcome on purpose. It's simply what happens when a scoreboard rewards attention instead of wellbeing.

The playbook for capturing a child's attention is already optimized in retail and digital. What is missing is a playbook for measuring whether the attention was worth capturing.

If you are building a product that lives in someone's home, the question is not whether it holds attention. It is whether the family is better off for having it there.

When the computer walks, talks, has a voice and a face, the opportunity is far greater than what it can do by command. What if kindness along with empathy were core KPIs? Designing for empathy means building machines that focus on the human experience. These should calm a frantic environment, assist communication between parent and child, and foster confidence. The opportunity is moving from being "users" of a device to "partners" in our own home. We want the machine around because it makes our lives better, not because it captures our time.

​
A New Scoreboard, Built Around Joy, Laughter, and Love

This is where the real opportunity sits, and it's bigger than children's products. As AI becomes an interface genuinely present in our daily lives, we have a rare chance to decide its KPIs before the old ones get copied over by default. Not session length, but joy. Not return frequency, but laughter. Not time on device, but love and connection strengthened rather than attention extracted.

These aren't soft, unmeasurable ideals. They're design principles. Did this family communicate better this week? Did this child grow a skill, a friendship, a piece of confidence? Did this home feel calmer instead of more fragmented? Outcomes such as these can be built for and measured the same way watch time and daily activities are today.

The next five years will decide whether family tech gets a different scoreboard or inherits the one that broke social media. Which one are you building?

The hidden costs of scaling decisions across borders.

I've sat in rooms in Kraków and Manila where this exact conversation was happening: a second site, built from the first site's playbook, somehow landed different numbers than anyone expected. The story below happened in India. But I've watched a version of it play out everywhere a capability center is asked to replicate itself in a hurry. The playbook is documented, the dashboard is shared, and leadership assumes that if the process travels, performance comes with it. It usually doesn't. And the reason it doesn't is something you won't find on that dashboard.

How GCCs Went From Running Tasks to Owning Decisions

A decade ago, a capability center's job was narrow. Take a defined process and run it more cheaply and cleanly than headquarters could. You could write that down. Hand it over. Audit it.

That's not what most GCCs do anymore. They own outcomes now, including product calls, customer escalations, and P&L for whole functions. Which means the thing being scaled has quietly changed from a task to a decision. A task can be handed over completely. A decision can't. Whoever makes it also needs the standing to have it stick, and standing is the part nobody thinks to package.

Why the Playbook Worked Once and Broke the Second Time

I watched a team hit every milestone on its roadmap nine months ahead of the global template. Leadership flew in to see it. The case study slides were half-written before the results had even settled.

Here's what nobody said out loud during the celebration. That speed wasn't really about process. It came from a hundred small, unglamorous calls the team had made over months, which escalation to override, which customer request to say no to, which metric to protect even when it made the weekly numbers look worse. Nobody wrote any of that down. Why would they? None of it looked like a step. It was a track record building quietly, one unremarkable decision at a time. You only notice it's missing once you try to hand it to someone who hasn't made a single one of those calls yet.

So the organization did what organizations do with things that work: they ran the same playbook in a new geography and expected the same speed. What actually got exported was a set of steps, minus the years of small trust that had made those steps cheap to take. The new team got the document. They didn't get the benefit of the doubt. Every close call had to be argued for now, because nobody upstream had any reason to take their word for it yet.

That's the real story behind the org-chart mess that followed. Nobody could answer the simplest question: What is this, organizationally? A new business unit? Does it fold into the existing one because the customers overlap? For a while, someone parked it under the CFO as an infrastructure project. As if the open question were square footage and server racks and not a team nobody had gotten around to trusting yet. A year in, that unit was answering to more than half a dozen reporting lines. Each applied its own scrutiny, because none had the standing to simply defer. Nothing got established. Nothing has been given clear ownership, and you cannot own a decision you don't yet have the authority to make stick.

Process is the kind of knowledge you can write down. It moves at the speed of a document. What that team had built was not written-down knowledge. It moves at the speed of a relationship, which is to say it doesn't move at all. Every site has to build its own, from nothing.

What GCC Leaders Who Scale Well Do Differently

The GCC leaders I've seen handle this well, don't treat a second site as a copy of the first. They treat earning that same standing as a separate job. Most organizations leave it to chance. These ones don't.

A few things they get right. Someone is named accountable for the new unit before it opens, not after the first unresolved call forces the question. Not a committee. One person, whose job explicitly includes soaking up the ambiguity the org chart hasn't caught up to yet. Often there's someone from the first site, or from global leadership, who spends the early months quietly vouching for the new team's calls. Lending them credibility before they've had time to build their own. Think of a co-signer for someone with no credit history. Same idea. And the new site gets judged on a different clock. Not whether it hits the same milestones at the same pace, but whether it's accumulating the kind of decision history that will let it move that fast on its own eighteen months from now.

None of that shows up in a steering-committee deck. It's not a process improvement. It's a choice. It means spending deliberate effort building something most organizations just assume will appear on its own. The presumption that a team's judgment is worth trusting before it has had the time to prove it.

The centers that scale well aren't the ones with the tightest playbook. They're the ones who understood, before the second site even opened, that they weren't exporting a process at all. They were asking a new team to spend authority that nobody had given them the chance to earn.

​

Choose People. Every Time.

Few OTT shows bring the whole family together and spark conversation the next day. Last weekend, we watched Made in India: A Titan Story on Amazon Prime. My parents were interested from the start. Even my six-year-old told his friend about it: You know this guy is the boss. They make watches. That’s the best summary I’ve heard.

Their enthusiasm made me realize the show isn’t just about watches; it’s about something much bigger.

It’s about building something despite doubt. Xerxes Desai and his team weren’t sure it would work, but they believed in what they were doing. Certainty waits for proof; conviction acts before proof exists.

What really stuck with me was how the show talked about people. At one point, there’s a choice between focusing on people or on process. The answer was clear: always choose people. It’s not that the process isn’t important, but if you build a process on distrust, you end up causing the very failure you wanted to avoid.

Choosing people isn’t always easy. It means being willing to accept mistakes along the way. The show doesn’t sugarcoat this. It shows the real cost, a wrong decision, a missed target, a product that doesn’t work out. Desai accepted that. He didn’t see every mistake as proof he’d trusted the wrong person. Instead, he saw it as the cost of building a team that would eventually succeed. That’s a different kind of leadership. Most organizations claim to value learning from failure, but few make it possible.

The bond between Desai and J.R.D. Tata is inspiring too. It’s about one person trusting another’s instincts before the numbers made sense. That kind of belief is harder to find than any strategy. Most of us can remember one person who believed in us like that. Not many of us have had more than one.

And then there’s how people in the show talk about Tata. It’s more than just loyalty; it’s almost like love. That comes from years of putting people first, time and again. This kind of people-first culture stands out even more today, when…

With so much news about layoffs and talk of AI replacing people, it’s refreshing to see a show that still believes in what people can do.

I was still thinking about it at my desk on Monday, and that doesn’t happen often.

If you haven’t seen it yet, try to watch it this weekend.

My corporate career was going well. That was the problem. 

I had spent more than a decade in digital media across India, Singapore, and the APAC region. The career gave me stability, growth, and identity. But over time, while I was progressing externally, I was no longer connected to the direction I was heading.

It was not dissatisfaction. It was a misalignment.

The question that kept surfacing was not “should I leave?” It was quieter than that: if I wait until retirement to find out whether this work could become something real, will I be able to live with that answer? I could not.

If your career looks fine on paper but inside it doesn’t feel right, you know what I mean.

From Personal Practice to Purpose

Around the time my role in Singapore became redundant, I had started exploring different wellness practices, including running and yoga. What began as a way to use that transition period productively led me to complete my first yoga teacher training.

The more I practiced and taught, the more connected I felt to the work. I found meaning in the smaller moments: adults leaving calmer, children learning to notice their breath from a young age, and families using simple tools to bring more ease into daily life.

After a few months, I returned to a corporate role. The job gave me stability, but it no longer gave me the same meaning as teaching yoga and mindfulness.

From 2015 to 2023, I continued building this work alongside my full-time career and family life. Most weekends I was teaching, learning, creating programs, or working with adults and children. It never felt heavy because I enjoyed it.

Eventually, I specialized in children’s yoga and mindfulness. Working with children taught me that supporting them requires more than teaching movement. It requires understanding behavior, emotions, attention, creativity, and connection.

As interest grew among parents, schools, and communities, I saw this work was more than teaching yoga on the mat. It was about helping children build self-regulation, confidence, resilience, self-awareness, and meaningful connections.

Passion and Purpose Alone Are Not Enough

One of the biggest factors that helped me move from interest to impact was consistency.

For years, I kept building alongside my corporate career. I did not leave immediately because I had not yet created a viable model that could support both the mission and the practical realities of life.

I started turning my side project into a sustainable platform. Physical classes could only reach so many people, so I asked: if I wanted to have more impact, how could this work go beyond live sessions?

That shift led me to build digital resources, online programs, flagship products, teacher-training certifications, and a broader vision for children’s well-being. I developed a digital kids’ wellness platform, launched my flagship Kids Yoga Teacher Training certification, expanded my work with communities, parents, educators, and families, and deepened my focus on special-needs yoga and inclusive well-being.

What began as individual classes gradually evolved into an ecosystem.

This is where many passion-led ventures face their real test. Passion creates momentum, but sustainable growth needs structure, systems, adaptability, and a long-term vision. This was not only a career transition; it was turning a passion project into a purpose-led business with structure, patience, and consistency.

The Moment I Had to Choose

For years, I balanced two parallel careers: my corporate role and the wellness work I had been building alongside it.

A few questions kept resurfacing. Would I be financially secure without corporate income? Why leave when my career is progressing well? Would I regret not giving my wellness work a genuine opportunity?

The turning point came when changes in my role and team structure pulled my energy away from the work I had spent nearly a decade building. Rather than making an impulsive decision, I took a sabbatical.

That pause gave me space to reflect. Could I commit fully to the work I had built? Was waiting until retirement really better? Would I regret not exploring this path?

The answer became clear. There would never be a completely risk-free moment. The difference was that I had the energy, experience, and momentum to pursue it now.

By then, BlissYogaSG was no longer a passion project. It has evolved into children’s yoga programs, teacher training, school partnerships, inclusive wellness offerings and a digital learning platform.

Build With Courage, But Also With Grounding

Today, when people ask me how to leave a job and follow their passion, I do not tell them to be impulsive. I tell them to start small, while still working, and let consistency give them clarity. Build a structure around the passion. Plan practically, especially around finances and responsibilities. Keep learning and evolving. And take the leap only when your inner clarity and your outer foundation begin to meet.

That moment will not feel completely safe. But it will feel real.

Today, my professional background continues to support my business. BlissYogaSG has grown into a wider ecosystem, with 500+ students certified worldwide as kids yoga and mindfulness teachers, globally recognised Yoga Alliance-accredited teacher training certifications specialising in children’s well-being, a digital-first children’s wellness platform with 60+ pre-recorded programs, and an upcoming Special Needs Kids Yoga Teacher Training. 

My work in children’s and family well-being has taught me that real change often begins with adults. When we support parents, educators, and teachers, they are better able to support children.

That is the work I chose, or maybe the work chose me.

Looking back, it does not feel like I left one career for another. It feels like I slowly built the courage, structure, and clarity to come home to the work I was meant to do.

​

The Question Behind the Headline

Meta laid off around 8,000 employees and moved 7,000 others into AI-focused roles.

The layoffs got a lot of attention, but I kept thinking about the employees who were reassigned. What are they being asked to do? They are now helping to build systems that could one day decide how much of their own work remains relevant. That is not a small ask. We have seen this kind of workforce shift, often described as an opportunity or a transition.

Lindsay Jessup, CEO of Geeks Ltd, wrote something for us worth reading twice. She said your AI works fine. The real bottleneck is decision-making. Figuring out which problems matter, which tradeoffs to accept, and which direction to take when several options seem equally valid. That is still human work. But it is harder than most organizations have prepared their people for.

Richard Hua, founder & CEO of EPIQ Leadership Group, observed in his article that even skilled teams can struggle with major change, and that the reason is usually cultural. When things feel uncertain, people look to their leaders. If leaders demand speed but react harshly to mistakes, people hesitate. Teams hold back precisely when they need to move.

Both observations point to the same problem. Organizations are asking people to move faster while creating conditions that make movement feel dangerous.

Somayeh Aghnia, co-founder of the London School of Innovation, raised the deeper question. Most organizations think about AI in two stages: what it saves now, and what new value it creates next. But there is a third important stage. When AI can do more of what the organization was built to do, the question shifts from efficiency to purpose. Why does this organization exist?

That question is always present in workforce change.

Most workforce announcements tell you who is staying and who is leaving. Very few tell you what the organization is actually becoming.

If you’re leading through these changes, what are you asking your team to build? And how are you protecting trust as they do it?

Ashwini has been running Kaara for long enough to know what a wrong bet feels like. He has made several. He no longer calls them failures.

What he carried out of those years was not a pivot strategy. It was something simpler: stay honest to the basics, follow the business case, and never fall in love with the idea.

The firm today has 220 people. It is growing while TCS and Oracle are cutting. Three months after Kaara.Code went public, JPMC called. He came back from that meeting thinking about a question he had not asked before: how do you give sixty-five thousand engineers the right blueprint and memory for their enterprise development journey? He has a clear position on what is happening in his industry, and it is more contrarian than most people in it are willing to say out loud.
This is that conversation.

​
Kaara has been through significant reinvention. What was the moment you found the clarity that formed what it is today?

We realized the shift in late 2024. The foundational models were getting better, our conversations with customers began to shift, Gen AI adoption took center stage, and soon the conversation moved to Agentic AI use cases and adoption. At the same time, we had started using AI inside our own delivery, in how we write code. There was real internal friction about it, and that friction turned out to be the most useful thing that happened to us.

The more we used AI, the deeper our conviction grew in the importance of context and memory for the proper adoption of Agentic workflows.

That was the moment it dawned upon us: if we wanted to help our customers realize what this shift makes possible, we had to change how we deliver before we asked them to change anything. That’s how the concept of Compounding Build started to take shape.

Before Kaara became what it is today, you tried Social, Edtech, Logistics, and Analytics. What did those experiences teach you about developing something that lasts?

We failed in all of those. None of them became what we hoped. I have moved on, and so has Kaara. We have stopped calling it failure. It was a learning experience, albeit a costly one. Costly enough to be worth remembering. Here is what I took from it.

Never ever fall in love with the idea. Follow the business case: why is it needed, who is going to use it, and how does it make their life better? Ideas are not enough. The discipline of execution, staying close enough to the ground, being able and willing to argue against your own plans, that is what actually matters. Cash flow is one of the most important aspects of any business, and founders often ignore it. The wrong culture can kill great ideas, and no execution SOP can save it.

What I learned is that building something that lasts is actually a lot simpler than most of us think. Today, I try to be honest in what I do, why I do it, who I do it with, and how I do it. Staying honest to the basics is something that I have started to value a lot.

​
The IT services industry is experiencing its most significant disruption in 30 years. TCS. Oracle. Major cuts. Kaara is still growing. How do you read the current situation?

Our industry is seeing one of the most challenging times in the last 30 years; it was bound to happen sooner or later. AI has not caused it; it just brought it to the surface. This is the unwinding of a pricing model that remained unchanged for 3 decades, the confusion about AI’s value, and enterprises hiding their past decisions in the guise of AI productivity. I am not sure which one is contributing how much. One thing I am 100% certain of is that it’s not machines replacing humans.

It is a mismatch between how the industry has long created value and how value should be created now. We are a small firm, from whichever lens you look at - an IT services company of 220 odd people is a small company, and there are always opportunities for companies our size, no matter the market sentiments. We also made a few bets early, which are paying well. And that’s where the opportunities are: we have to reimagine how we deliver value to our customers, enable our team, and move them up. The days of the business model that assumed scale would always equal growth are long gone, but that’s also where the real opportunities lie.

Kaara.Code was recognized at JPMC's Technology Innovation Forum. What did you learn from that experience?

It has been around three months since we started talking about Kaara. Code and Compounding Build publicly. It was a pleasant surprise to receive a call from JPMC to present under the Modern Engineering Practice track.

The heartening part was understanding what they are doing internally. The context engineering work they are attempting is what Kaara.Code provides systematically. Before meeting them, we didn't realize how important it was to answer the question of scale. JPMC has a tech team of more than sixty-five thousand people. We came back thinking about how we can give each of them the right blueprint and memory for their enterprise development journey. That question changed the size of what we were building.

You have written that people building software matter more in the AI age, not less. That is a contrarian position right now. What gives you the confidence to hold it?

The common assumption is that as AI gets better at writing code, the people who write it matter less. I believe the opposite. My confidence comes from seeing the way Kaara functions today.

It's true that most developers haven't written a single line of code in months. But coding is not the only aspect of building software.

Knowing what is being built, why it is being built, end-user personas, context, compliance, guardrails- all of that has become much more important than before. The role of developers and builders has become more important too, but it moves upstream. It moves to judgment, knowing what to build, which trade-offs to encode, which constraints are real, and what 'done' actually means for a business user.

Furthermore, we are now solving problems that were hitherto unsolvable. The more we build, the more we grow, and the more work there is for everyone. We need to move into this cycle fast.

What is the one thing most companies miss when they move into AI, and why does it keep happening?

Stop inserting AI into old workflows. This is a sheer waste of such a powerful technology, and a lot of us keep taking the easier path.

The easier path is to retrofit AI into existing ways of functioning. What is actually required is to reimagine the work itself from the ground up. The way enterprise services are delivered needs to be completely reimagined. The roles of developers and everyone involved have changed; that does not mean they are going away. It is changing, that is it.

The best part: almost everyone I speak to is realizing this now and ensuring reimagination of workflows in almost every new project implementation.

My simple view is this: The winning companies will be the ones that redesign work, rebuild governance, make AI part of the business's daily operating rhythm, and do it fast.

​

Electrification is scaling manufacturing fast, but operational excellence still needs a rethink.

The quality engineer spent about 40 seconds at our monitoring screen before he asked his question. He didn't ask about our defect rate. He asked: "What's your process capability on this dimension across all three shifts?"

I knew the average. I didn't know the split by shift.

Over decades of manufacturing, I've sat through more customer audits than I can count at Bosch, Gabriel India, Auto Ignition, and now Endurance. Nobody had ever asked me that. That question was the first clear signal that electrification wasn't just changing the components we make. It was changing the standard by which our manufacturing is judged.

This made me realize that bigger changes were underway in how we are measured. That realization set the stage for everything that followed.

How Electrification Is Changing The Shop Floor Reality

Electrification has changed what good manufacturing means. Electric and hybrid parts need to be made with more accuracy and consistency than most traditional car parts. At Endurance, where we make brakes and suspension for companies building electric vehicles, we've had to rethink what "ready for production" means. There is less room for error, and we must be able to track every part.

The technology response was predictable and, generally speaking, correct. In-line quality monitoring, real-time process dashboards, and predictive maintenance have moved from pilot projects to core capabilities. Manufacturing systems are becoming more digitally connected, sensors and machine data allowing faster detection of variation, shifting teams from reactive problem-solving to preventive and predictive control. Plants are being redesigned with more flexible layouts, a higher automation density, and built-in traceability.

I didn't expect that being good for the environment would become so important for how well we run. Watching our energy use, using materials wisely, and making less waste are now key to staying competitive. These things are not just for CSR reports; they are part of how we work in every electric vehicle project.

The scale of this shift is visible across global manufacturing. From hubs in India to plants in Eastern Europe and Latin America, electric mobility programs are compressing development cycles, increasing quality expectations, and demanding higher levels of process stability than most operations built for ICE were designed to deliver. But as fast as these changes are happening worldwide, not every challenge is being openly discussed.

The Gap That Nobody Wants To Talk About

Even though we use more machines and digital tools, results are often not steady. I've seen this in my company and others. The problem is not what people can do, but how well they follow good routines. If the way we work is not strong, technology only shows us the problems but doesn't solve them.

There is also a skills gap. Making electric vehicles means learning new materials, safety rules, and quality standards that weren't needed before. Workers and engineers need to know both how things are made and how to use digital tools. This mix of skills is rare and takes real effort to build.

We had real-time alerts. We had dashboards. What we didn't have was a standardized response at the supervisor level across all three shifts. Problem-solving was personality-dependent; it worked when the right person was on the floor, but not when they weren't. The data was telling us everything. The team wasn't consistently acting on it.

What Actually Scales

The next phase of manufacturing excellence will be driven by integrated operating systems, shared performance frameworks, and standardized ways of working across plants and geographies, because inconsistency at scale is expensive, and electrification makes the cost visible faster than any previous transition.

People's skills will set companies apart. Being excellent comes from people working well together, helped by technology, not the other way around. The best companies will understand this and invest in their people.

Plants must be built to handle problems and bounce back. New technology helps us move faster, but true success comes from putting technology, strong work habits, and people development together, in that order.

The Question Worth Sitting With

If you're leading a manufacturing operation through this transition, the question of technology investment is probably already settled. The harder question is whether your operating model can absorb those investments and turn them into consistent output.

Your sensors will tell you everything. The important thing is what your team does with that information.

What does your plant look like at 2 AM when something goes out of control, and the right engineer isn't there? If you're not certain of the answer, that's where the real work starts.

​

The Right Book at the Right Time

Two books are on my bedside table right now. Indra Nooyi's My Life in Full and Adele Faber's Siblings Without Rivalry. One is about leading one of the world's largest companies. The other is about managing two children at dinner. Both feel urgent.

That is the thing about books. The right one at the right time does not just give you information. It gives you language for something you are already living through. The wrong one, even a brilliant one, slides off completely.

That is exactly why we started Reframe and Read. Not a list of books every leader should read. A specific book for a specific leadership problem, at the moment the problem is live.

Five issues in. Here is what we have covered.

What's Your Problem? by Thomas Wedell-Wedellsborg
The problem: You are trying to solve the wrong problem. The first issue your team identifies is almost never the real one, and most organizations have no way to verify it.

The Anxious Achiever by Morra Aarons-Mele
The problem: You are holding back your anxiety before every important meeting. Acting calm has a cost it blocks the signal that shows where the real risk is.

Necessary Endings by Dr. Henry Cloud
The problem: You are avoiding the ending that must happen. Most reorganizations fail not because the new setup was wrong, but because the leader could not stop what needed to be stopped.

Decisive by Chip Heath & Dan Heath
The problem: You think you are making logical decisions. Research shows this is not true. Our brains are too confident, look for information that supports what we already think, and get taken over by short-term feelings. Knowing about these biases does not fix them; you need a completely different approach.

Multipliers by Liz Wiseman
The problem: You think you are helping your team by staying involved in the thinking. Liz Wiseman spent years studying what makes leaders bring out the best in others or shut them down. It depends on whether a leader sees their job as doing the thinking or making it possible for others to do it.

The right book does not give you answers. It gives you the right question at the right time. That is more valuable than reading many books at the wrong time.

What are you reading right now, and what problem is it solving ?

Read. Reflect. Share. We are just getting started.

Your AI works fine. Your decision-making is the bottleneck.

Four boardrooms. Four weeks. London, Texas, Singapore, France. I’ve seen a logistics CEO worried about warehouse automation, a hospital group rethinking clinical processes, a retailer overwhelmed by AI-generated forecasts, and a manufacturer asking what its supply chain will look like in five years.

Different industries, identical conversation: everyone claims to have AI now, but I rarely see anyone making the most of what it produces.

That's the dirty secret of the AI age: the technology works. The humans haven't caught up. Your team can generate strategy decks, clinical pathways, financial models, and production code before lunch. There is more to sift through, and less time to figure out what matters.

AI amplifies whatever you put in. With sound judgment and data, returns grow. With bad judgment, mistakes escalate. It all happens faster and on a bigger scale, with confidence.

This isn't an intelligence problem. It's a judgment problem. And judgment that doesn't turn into decisions, priorities, and outcomes is just expensive thinking.

Speed Went Up. Clarity Didn't.

For two decades, leadership wisdom focused on speeding things up, automating, and scaling. It worked. Until it didn't.

So what does this mean for leaders? AI changed what leadership requires. When your team can produce in an afternoon what used to take two weeks, the hard part isn't production. It's knowing which output to trust, which to bin, and what to do next.

Without people who can discern signals from noise, every bad assumption compounds. The dashboards look sharp. The team sounds sure. Meanwhile, errors scale alongside the outputs that are costing you more than ever before.

The Accountability Problem Nobody Wants To Talk About

AI is brilliant at producing plausible responses. It synthesizes, summarizes, recommends, writes code, and drafts contracts. Blank page to first draft in minutes.

But most companies still talk about AI like it's just another tool, another technical advancement for IT to sort out. It's not. AI is an intelligence. Tools don't make decisions; they follow rules. Intelligences do. And that changes everything about oversight, accountability, and integration.

The org chart shows who's responsible, but no one's mapped how AI flows through technical workflows. Who owns the output when the machine wrote the first draft? Who catches the error when the model hallucinates? Who's accountable for setting guardrails so an AI-assisted decision doesn’t go astray?

Most companies have bolted new intelligence onto old structures and hoped for the best. That's human work to fix. Leaders who can hold complexity, break it down, prioritize what matters, and adapt when the ground shifts. Those are the people who need to own the accountability gap, not the vendors, not the model, and not IT.

Better Questions, Better Decisions

In each boardroom, I noticed a pattern among the most effective leaders. They were the ones asking the toughest questions. They didn’t just wonder whether the AI got it right; they asked how they’d know if it went wrong and how quickly they could adapt. Rather than accept the model’s recommendations without question, they wanted to know who would bear the cost if it failed, and whether those people had truly been heard.

Sometimes the right call is an extra step or a human checkpoint until the models advance. The leaders who get this right know when to get the best from AI, when to challenge it, and when to ignore it entirely.

I call these people Chaos Cartographers. Not because they predict the future. Because they steer uncertainty when it's still moving.

Judgment Is Infrastructure Now

Most AI strategies are still tool-obsessed: pilots, governance, prompt libraries, productivity dashboards. Fine. Not enough.

Executive judgment is what's missing: how your leaders break down problems, decide, prioritize, deliver, and adapt when things move faster than the plan. Find those people. Develop them. Put them where the hard calls get made.

Because transformation doesn't run on instructions. It runs on conviction. People follow leaders through times of uncertainty. They don't follow dashboards.

 
The Map Is Not The Territory

AI generates maps endlessly. Beautiful, confident, formatted maps. All plausible. Many wrong in ways you won't spot until later. The map is not the territory. The dashboard is not the business. The output is not the outcome.

In the AI age, you don't need to be the smartest person in the room. You need to be the one who can navigate when the room is full of intelligence but still short on wisdom.

Better tools matter. Better cartographers matter more.

The next divide won't be AI versus no AI. It'll be companies that use AI to compound their returns, rather than those that compound their errors.

​

Why AI Gives You the Tools But Not the Judgment to Use Them

Giving everyone a paintbrush does not make everyone Picasso.
That is where we are with AI right now. Across media, adtech, and content, the tools are available to almost anyone. The barriers to access have genuinely collapsed.

A publisher in Lagos, a programmatic trader in Jakarta, a content team in Bengaluru: all of them are running on roughly the same AI infrastructure as organizations ten times their size. That is a remarkable thing. It is also, in my experience, where the confusion starts.

Access was never what separated good from great in this industry. What separated them was knowing what to do with it.

AI Moved From Backroom To Boardroom Faster Than Anyone Expected

Three years ago, AI in advertising and media was largely an infrastructure story: smarter bidding algorithms, better fraud detection, faster content tagging. It sat inside the tech stack, mostly invisible to commercial leaders.

That changed. Today, AI touches every layer of how content is made, how consumers are understood, and how inventory is priced and sold. Generative models can produce video and text at a regional scale in dozens of languages. Programmatic platforms have embedded predictive yield layers that can recommend floor prices in real time. AI-driven audience segmentation now runs on first-party signals that were unimaginable in the cookie era.

The shift isn't incremental. In India, publishers who were manually curating regional content a few years ago are now running AI-native studios that produce thousands of pieces daily, each localized for dialect, format, and device. In Southeast Asia, adtech platforms are using AI to match small advertisers with hyper-local inventory that no human sales team could have managed at that scale.

The democratization is real. AI genuinely lowered the barriers to content creation, programmatic access, and audience insight. A mid-sized publisher in Lagos or Manila now has access to tools that would have required a room full of engineers in 2018.

That is the shift. More capability, more broadly distributed, faster than the industry expected.

More Signals. Not Necessarily More Wisdom.

When we started building video content at scale using AI, I assumed the tools would do most of the heavy lifting. Feed in a brief, get a video out. That is not how it works.

To get anything usable, you need to know what a good script looks like before you prompt for one. You need to understand storyboarding well enough to catch when the visual logic breaks. You need to have an instinct for camera movement, voice texture, image composition: not to do those things yourself, but to recognize when the AI has got them wrong. The output is only as good as the craft knowledge you bring to the input. Without that, you get volume. You do not get content.

The same is true in programmatic. AI can optimize a bid, enrich an audience signal, and recommend a floor price in real time. What it cannot do is walk into a room and make a CMO trust that their brand belongs in your inventory. The biggest deals I have worked on did not close because of a dashboard. They closed because of a conversation that happened before the dashboard was ever opened. AI had no role in that room.

AI can harness your talent. It cannot replace it.

This is the gap that does not show up in the case studies. The tools amplify what you already know. If you do not know enough, they amplify the gap instead.

The Practitioners Who Combine Both Will Set The Pace

Knowing the gap exists is not enough. The question is what you do with that knowledge. The practitioners who are pulling ahead are doing three things differently.

Content: From Volume to Value Architecture
The next phase of AI in content is not about producing more. It is about producing smarter, with a clearer link between what gets made and what gets monetized. The publishers who will win are those who use AI to understand which formats, which topics, and which audience signals drive real commercial outcomes, and then use human editorial judgment to build content supply around those insights. AI helps you see the pattern. Humans decide what to do with it.

Programmatic: From Automation to Contextual Intelligence
The best adtech professionals are starting to think about AI not solely as a bidding tool but as a context-enrichment layer. The question is shifting from 'how do I automate the transaction?' to 'how do I make every impression more defensible to a premium buyer?' AI can enrich inventory signals, map content quality to advertiser outcomes, and identify yield gaps in real time. The human role becomes one of packaging intelligence into commercial narratives that advertisers actually buy. That is a skill, not a feature.

Partnerships: The Relationship Layer Stays Human
Across both content and adtech, the commercial partnership layer remains unwaveringly human. The most sophisticated AI systems on the market cannot negotiate a content syndication deal, cannot build the mutual trust that makes a first-party data partnership work, or walk into a room and convince a CFO that their brand deserves to be in a different conversation. These are judgment calls, built on experience and credibility, and they will stay that way for the foreseeable future.

The professional who understands craft will set the price. Everyone else will compete on volume.

The Paintbrush Is Not The Point

Every professional in media, adtech, and digital media now has access to the same palette of AI tools. That access is real, and it matters. But access was never the constraint that separated good from great in this industry.

What separated good from great was knowing what to paint, for whom, and why it mattered. AI does not answer those questions. It just gives you more ways to answer them yourself.

The professionals who treat AI as a productivity amplifier rather than a judgment substitute will produce work that is faster, richer, and more commercially grounded than anything they could have done alone. The ones who treat it as a shortcut will produce a lot of output that lands nowhere.

The paintbrush is not the point. The artist is.

​