Eight years ago, I made a hiring decision that didn’t follow the usual rules. In today’s AI era, where tools can already assist with much of the work that we do, that decision feels even more relevant in hindsight. I was hiring for a fast-growing fintech company. The shortlist was what you’d expect: candidates from top universities, strong academic records, and big brand internships. On paper, most looked solid.

But one candidate stood out for a very different reason. He didn’t have a strong degree. But when we gave him a problem to solve, something clicked. He broke it down clearly, asked the right questions, and explained his thinking in a simple, structured way. He was fast, practical, and very focused on outcomes.

Our HR process didn’t allow us to hire him directly. We had to get special approval from leadership. We hired him anyway.

Within a few weeks, it was clear we had made the right call. He delivered faster than many experienced hires, not because he “knew more,” but because he could think clearly and adapt quickly. That moment changed how I look at talent.

What AI is changing right now in hiring

Today, that experience feels even more relevant. AI is now doing a lot of the work that used to help candidates stand out, right from basic research, writing first drafts, building summaries, and even structuring analysis.

We are seeing these changes across industries and, more importantly, across our own client ecosystem, where AI adoption is already becoming operational, not experimental.

For instance, BASIC Home Loan (a leading fintech) uses AI to simplify home loan journeys and reduce friction in early-stage processing, while KhiladiPro (sports tech startup), uses AI to make sense of large volumes of sports and engagement data. Both cases reflect something important: our clients are not just using AI; they are becoming AI-native in how they redesign workflows. And that changes what we value in people.

Also Read: What to actually prioritise when your board wants AI and everything feels urgent

This is not just a company-level shift. It is showing up in the broader labour market as well. The India Skills Report 2026 reinforces this transition: India’s employability stands at 56.35 per cent, even as demand rises sharply for digital and AI-related skills. India contributes about 16 per cent of global AI talent, positioning it as a key global supply hub; and by 2027, the country is expected to have 1.25 million AI specialists.

Taken together, the signal is clear: AI is not just changing tools inside companies, it’s changing how work is structured, and how talent is defined.

So what does a future-ready skill stack look like?

The hiring question has shifted. It is no longer: “Who has the best profile?” It is now: “Who can use new tools to think better and work faster?” Across teams, three shifts are becoming very clear—skills matter less than learning speed, thinking has become the real edge, and execution is now human + AI rather than either/or.

If that is how work is changing, the skill stack is also changing—from isolated abilities to a few connected strengths that work together.

• Industry knowledge + AI fluency: Understanding your sector’s economics, policy shifts, and blind spots is still important, but it now works best with strong AI fluency. The real edge is knowing how to use AI to speed up work, improve decisions, and sharpen thinking—not just to generate output.
• Thinking, data and storytelling: AI can process data, but it cannot decide what matters. The ability to find signals in data, simplify complexity, and turn it into clear, decision-ready storytelling is what helps people influence outcomes.
• Collaboration + communication: As more work becomes AI-assisted, human skills matter more in how teams align and move together. The ability to build trust, persuade, and coordinate across people is becoming a key leadership skill.
• Creativity: Automation improves efficiency, but it does not replace originality. In fact, as execution becomes easier, the ability to think differently becomes even more valuable.

Also Read: Think with AI: The new skill for social entrepreneurs

If I were hiring today

I would still make the same decision I made eight years ago. But I would add one more filter. Not “Which college?” Not “Which degree?”

Instead, I’d ask: Can this person learn continuously? Can they adapt as technology changes? Can they use AI to improve how they think and solve problems? Can they combine human judgment with machine capability? Because that is increasingly the real job.

Final word

The biggest talent shift in the AI era is not that degrees matter less. It’s that learning speed, adaptability, curiosity, and AI fluency now matter more than ever before. The strongest teams of the future will not simply be defined by credentials or polished resumes. They will be defined by how quickly they can learn, evolve, and work alongside AI to create meaningful outcomes.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post What hiring a high school graduate taught me about talent in the AI economy appeared first on e27.

About a year ago, I started noticing a pattern in how clients were briefing us. They don’t just come to us for deliverables anymore. They were coming with questions. Which AI tools should we be using? Should we be using them at all? What does a responsible AI-assisted creative process even look like?

I spent seven years in government service before founding CreativesAtWork 14 years ago. Across those two very different worlds, I have sat across from a lot of different kinds of clients. But this was new. They were not just buying a service. They were looking for someone to think alongside them — someone who understood both the craft and the tooling, and could help them make sense of a landscape moving faster than anyone expected.

That shift told me something important: the gap isn’t just between AI and creatives. It’s between what the market now expects and what our talent pipeline is actually producing.

What clients are actually asking for now

The ask has quietly changed. Clients today don’t just want execution. They want judgment. They want someone who can look at an AI-generated concept and tell them honestly whether it’s on-brand, culturally appropriate, or legally murky. They want a creative partner who knows when to use the tools and, just as importantly, when not to.

That is a fundamentally different profile from what most creative training — formal or informal — has historically built toward.

And yet the conversation in most education and training circles is still centred on outputs: learn the tools, build the portfolio, get the brief done. The advisory layer — the ability to help a client think through AI adoption in their own creative process — is barely on the curriculum and it should be.

Also Read: What to actually prioritise when your board wants AI and everything feels urgent

Where the pipeline is falling short

Across Southeast Asia’s creative economy, which includes a large and growing freelance workforce, this mismatch is structural. Universities producing design, communications, and media graduates are still largely running programmes built for a pre-generative-AI world.

Bootcamps have moved faster, but they are largely targeting engineers, marketers and tools-specific and focused. Few focused on the mindset shift and the design thinking process together with the client.

The freelance community is largely left to self-educate. Platforms like Coursera and LinkedIn Learning have added GenAI modules, but most of what exists treats AI as a feature to understand rather than a professional context to navigate.

What actually would help

Peer-led communities are filling the gap faster than institutions. Regional design communities, Discord servers, and creative Slack groups are where working freelancers are sharing real workflows, catching mistakes, and pressure-testing AI output against live briefs. The learning is messy and informal, but it is grounded in actual work, which is more than most formal programmes can say.

Agencies that have restructured around AI are also quietly becoming training grounds. When the workflow changes, junior people inside it develop judgment faster — because they are doing real work, not doing exercises. That will be increasingly important, and how a good creative training will work. The tools are new. The apprenticeship logic isn’t.

The structural barriers are slowing progress

Three things are holding this back at scale.

Curriculum cycles move too slowly. Degree programme updates will take a while. GenAI capabilities are shifting every six to twelve months. It needs to catch up faster.

Assessment needs to be caught up as well. It is still hard to tell whether someone can genuinely work well with AI or is just using it to cut corners. Without clearer signals of quality, the market will not be able to reward real capability development the way it should.

Also Read: Think with AI: The new skill for social entrepreneurs

And the freelance workforce has no institutional home. Salaried workers access employer-sponsored training. Freelancers largely is on their own. In Singapore, SkillsFuture credits help, but take-up among creative freelancers stays low — partly because the available courses aren’t yet matched to how creative AI work actually operates day to day.

What needs to change

I would think an actionable next step for creatives is to have an industry body to lead the development of a portable, competency-based credential for AI-assisted creative work — built with practising freelancers, not just academics. One that assesses judgment and workflow design, not just tool familiarity.

For business leaders, rewrite your briefs. If you need someone who can direct AI-assisted production and advise on which tools fit the project, say that clearly. Vague calls for “AI-literate creatives” produce vague results.

And for educators: spend time with freelancers who are actually doing this work. The curriculum you need already exists in practice. It just hasn’t been written down yet.

The creative economy has always run on a mix of formal training, peer learning, and hard-won experience. What’s changed is how fast the ground is moving underneath all three.

Twenty-one years across two very different sectors has taught me one thing consistently: the people who stay relevant are not the ones who master the tools best. They are the ones who never stop asking what the work is actually for. That skill no model generates. But we need to start teaching it now.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post The creative gap: Why GenAI is outpacing the talent it was meant to empower appeared first on e27.

There is a number buried in the Agentic Economy Report by blockchain firm Morph that should concern every payments executive in Asia: US$0.20. That is the average size of an x402 agent payment as of March 2026, according to on-chain analytics firm Lookonchain. It is also below the fixed per-transaction floor of every major card network on the planet.

Card rails were engineered for human-sized purchases. A US$0.20 payment — the kind an AI agent makes when it calls an API, queries a data source, or pays for a single unit of compute — is not economically viable on Visa or Mastercard. It is not that the networks choose to exclude this transaction category. It is their cost structure that makes it impossible to serve.

Also Read: The invisible shopper rewriting Asia’s e-commerce playbook

This is the core insight behind the Morph report’s Prediction 2: most agent-initiated payments, by transaction count, will settle outside traditional card rails.

The x402 protocol and Asia’s stablecoin moment

The mechanism enabling this shift is x402, an HTTP 402-based protocol for stablecoin micropayments at API call time. The standard was co-developed by Coinbase, Cloudflare, and the x402 Foundation, and Google’s AP2 added a native x402 extension for stablecoin settlement when it launched with 60-plus partners in September 2025. The protocol is elegant in its simplicity: when an agent requests a resource, the server responds with a 402 (Payment Required) status, the agent pays in stablecoin, and access is granted — all without a human card number, a payment gateway, or a merchant acquirer taking a slice.

The current volume is modest, but the trajectory is being watched carefully. Bloomberg, citing x402.org data, reported approximately US$24 million in 30-day x402 agent payment volume in early 2026. Andreessen Horowitz’s crypto research team placed the wash-trade-filtered figure at roughly US$1.6 million over the same window, a 15-fold gap that the Morph report acknowledges directly, noting that raw volume is not yet a reliable indicator of commerce. What is reliable, it argues, is the speed of protocol integration at major payments, retail, and platform companies.

For Southeast Asia and broader Asia Pacific, where cross-border payment friction remains high, and stablecoin adoption has grown fastest among the world’s unbanked and underbanked populations, the structural opportunity is substantial. Circle’s USDC Gateway is already built to batch thousands of micropayments before on-chain settlement, a design choice, the report notes, “that is only useful when machines, not people, are the senders.”

The agent overtakes the human

Prediction 3 in the Morph report is the most structurally significant for the stablecoin sector: AI agents will overtake humans in commercial stablecoin payments by transaction count. The claim is deliberately narrow; it is about machines paying for real goods, services, or API capacity, not human transfers or conventional trading bots.

Also Read: Stablecoins are becoming ‘dollars as a service’ for emerging markets

The case rests on two diverging trend lines. Human commercial stablecoin payments — people buying goods with USDC or USDT — are bottlenecked by merchant acceptance, retail onboarding, and regulatory uncertainty. They compound slowly, year over year, at a rate of low-to-mid millions of transactions globally. Agent commercial payments start from near zero and compound monthly. Three drivers make this credible.

MCP SDK installs grew from 5 million monthly in February 2025 to 97 million monthly by March 2026, putting agent tooling in the hands of virtually every enterprise developer worldwide. ERC-8004’s live deployment on the Ethereum mainnet provides agents with a portable identity and on-chain reputation not owned by any single platform, removing a key barrier to agent acceptance at the merchant layer. And the Bank for International Settlements, in Working Paper 1310, concluded that routine cash-management tasks could be automated using general-purpose large language models at institutional quality.

Tether’s CEO, Paolo Ardoino, has publicly set the endpoint at “one trillion AI agents” transacting on-chain within 15 years. The technical ceiling, the Morph report argues, is gone. “The commercial ramp is what is left.”

The institutional layer is the harder problem

For Asia’s banks, payment networks, and fintech infrastructure providers, the more pressing question is not whether stablecoins will carry machine-to-machine volume — that outcome is increasingly structurally inevitable — but whether they will build the institutional rails to handle it under compliance regimes.

JPMorgan’s Kinexys platform now processes US$5 to US$7 billion daily and has launched USD deposit tokens on Base. Project Agorá, co-ordinated by the BIS, is in testing with seven central banks on a tokenised, programmable cross-border rail. Fed Governor Christopher Waller delivered four public speeches on agentic AI in payments between September 2025 and February 2026.

Also Read: The growing adoption of Ethereum in emerging markets

The report quotes Richard Astle, VP Head of Network at Fireblocks, framing the institutional challenge precisely: “Letting an agent transact on behalf of a user, at scale, and under a real compliance regime is a question of how you delegate authority without delegating custody, how you enforce policy at agent speed, and how you reach end users without rebuilding the stack underneath them.”

For Singapore, which has positioned itself as Asia’s digital asset regulatory sandbox, and for the broader region’s central banks navigating their own stablecoin frameworks, this is the operative design question of the next 24 months. The volume is coming. The question is which rails it clears on.

The post The US$0.20 payment that could rewire Asia’s financial rails appeared first on e27.

Every CV in my inbox is a perfect fit. All keywords match. All important tools are listed. The bullet points read like the job description, but slightly rearranged.

Three years ago, this meant something. Now it tells me almost nothing.

The real interview starts when the candidate has to speak without a script. That’s where the gap keeps showing up: young people who can produce output but can’t defend or question it, and often can’t explain how they got there.

We’re not being honest about this part.

The AI-native hiring pitch

Everybody keeps saying: junior hires are AI natives. They are faster, cheaper, and more fluent with the tools. They’ll out-execute mid-level employees still doing things the old way. Hire young, hire AI-fluent, scale your startup with half the headcount.

As a startup founder, I see the appeal. Most of us are looking for places where AI can replace manual processes and where output can grow without payroll growing with it.

The hiring decision that follows from it is where things stop looking so straightforward.

What the data says

Younger candidates are more comfortable with AI on average. That’s true. They reach for it without thinking. They save real time on tedious work.

But being comfortable with a tool is not exactly the same as being good at your work.

A 2025 Microsoft and Carnegie Mellon study surveyed 319 knowledge workers and found that the more someone trusted GenAI to handle a task, the less critical thinking they brought to it. Gerlich’s research on 666 participants showed that people aged 17 to 25 had the highest AI dependence and the lowest critical thinking scores of any age group. In engineering, a recent analysis found that junior developers accept around 89 per cent of AI-generated suggestions without seriously reviewing them.

Also Read: What to actually prioritise when your board wants AI and everything feels urgent

There’s something deeper going on underneath those numbers. In a Harvard Business Review piece from February, consulting partner David Duncan notes that generative AI was helping him much more than it was helping his junior colleagues. People with experience can tell whether AI output is good. People without it can’t.

So the question isn’t whether juniors are bad at using AI. They’re not. The question is what happened to the work that used to make them good at everything else.

Where junior judgment used to come from

The boring tasks that AI now handles, like first drafts, data cleaning, call summaries, and basic reports, weren’t just boring work. They were the apprenticeship. You learned what a good campaign brief looked like by writing fifty bad ones. You figured out which client signals mattered by sitting through the call and writing the notes yourself.

Researchers at IESE have started calling this the “skills pipeline” problem. Without the grunt work, it gets hard to imagine how anyone develops the expertise to step into senior roles. The same automation that boosts productivity this quarter can leave you without strong people to put in strategic positions three years from now.

So when I watch a junior confidently present AI-generated work, the question in my head is: where did your judgment come from? Sometimes there’s a good answer. Often there isn’t.

The other half of the gap

The flip side of the over-trusting junior is the under-adopting senior. In a startup, that one costs you just as much.

Senior people bring context, scepticism, and pattern recognition that catches AI errors. But a lot of them are slow to pick up the tools. Sometimes it’s a habit. Sometimes it’s that they’ve spent years being rewarded for doing things carefully and by hand. BCG’s 2025 AI at Work survey found that while three-quarters of leaders and managers use generative AI weekly, only 51 per cent of frontline employees do. The adoption gap runs across roles, not just generations.

A small company can’t afford either version of this. The junior who trusts AI too much and the senior who uses it too little have exactly what the other is missing.

Also Read: How AI is changing what an SME team actually looks like

Rebuilding the apprenticeship loop on purpose

If AI replaces the work that built judgment, hiring for judgment won’t be enough. You have to engineer it back in. In a startup with no patience for slow on-the-job learning, that means designing for it deliberately. Here are the four things to consider.

If AI ate the work that built judgment, hiring for judgment won’t be enough. You have to engineer it back in. Four things are worth trying.

• Score judgment, not output: Evaluate junior staff based on the questions they ask before execution, not the polish of what comes out. A junior who flags three problems with the delivery is worth more than one who turns in a clean draft on the first try. The World Economic Forum calls this the shift from execution to discernment.
• Pair juniors and seniors (both ways): Seniors review juniors for missed nuance. Juniors show seniors what the tools can already do. IDC and CIO Magazine now treat this as baseline practice, not a perk.
• Put juniors near the consequences early: Client calls from week one, not month six. As CTO Magazine put it, judgment forms through “imperfect decisions, confronting trade-offs, and experiencing feedback loops.” Without exposure, it doesn’t form at all.
• Make AI adoption a KPI for everyone: Expect seniors to use AI weekly, log where it fails them, and share those failures in team reviews. Seniors engage with the tools. Juniors see what informed scepticism actually looks like. Short internal hackathons, real workflow, mixed pairs. This makes work a “forcing function”.

What does this change about hiring?

“AI-fluent junior” as a category is incomplete. Fluency without judgment gives you confident-sounding work that can come out as faulty under review. Judgment without fluency gives you careful work that arrives a week too late.

What a small company needs is not a different kind of hire. We need a different setup for the hire. Stop looking for the candidate who already has both sets of skills – those profiles are very rare. Hire for the strength they have, and build what’s missing on the job.

The companies that work this out won’t be the ones with the most AI-fluent juniors. They’ll be the ones who still know how to grow a senior.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post Hiring an AI-fluent junior is easy, building one with judgment is the problem appeared first on e27.

AI tools may one day be better than any human at many things. It’s possible an AI system could become a better writer, artist and coder. The popular narrative is that humans will always dominate in roles where empathy, collaboration, and social interaction are important. We encourage people to spend more time on soft skills, collaboration and communication.

What if this isn’t true? In some settings, humans are reliant on AI tools for therapy and connection. Some humans prefer the company of animals to that of other humans. It seems reasonable that there will be a future where some humans prefer the company of robots to humans. We can already see signs of this through the dark-mirror-esque recent story from Wired about the wives of Silicon Valley founders. And all the reports of people seeking advice from chatbots rather than doctors or mental health professionals.

I can imagine a future where many people prefer a non-human system for many things that today we rely on humans for. We accept non-human decision-making in a wide range of industries and experiences today. Yet, just because AI systems and tools can do one or many things that humans do today better tomorrow, doesn’t automatically mean that humans will have no work to do. That is a false equivalence. There are an almost infinite number of possible futures where humans and automated systems work in different, varied ways.

Humans will remain relevant with or without AI tools because we employ people mainly to work with people, not to do mindless, automatable tasks. The majority of time in a typical job isn’t especially productive. People chat over coffee, handle personal matters, speak with customers, suppliers and partners and sit in endless meetings about things they should or could or might be doing. Productive, deep work does happen for most people, yet it’s often done around the other messy-humans-working-with-humans work. Even if you get a perfect answer every time, who’s going to build it with you? Once you have optimised your time beyond a certain point, there isn’t any more flex. We all need other people to get anything meaningful done.

Also Read: A 65% probability explains the next likely move for Bitcoin as leverage clears

Chatting with LLMs, chatbots, and agent tools can be interesting and fun, but it is also superficial. The longer someone chats with a chatbot, the more endearing the complexity of a human conversation becomes. Only humans can do many things with other humans. Only a human can have deep technical knowledge about a topic, have experienced that domain, and sit down to talk with you about their kids, their hobbies and complex problems. Only humans can create art that has a human story behind it. AI systems will certainly write amazing stories and generate spectacular art. But once it’s prolific, it’s worthless. We value things made by humans because of their scarcity, personality and story.

This is obvious to students and the majority of people who are increasingly negative about AI, yet it seems invisible to many leaders. I suspect this is because of the distance abstraction creates. Leaders of companies abstract themselves from the front-line work. Even a mid-sized company is simply too large to keep track of everything. Sometimes by choice or personality, managers increasingly abstract their connections with work and people. “It’s lonely at the top,” as the saying goes. All the potential possibility of AI systems fits nicely into this simplified abstraction. The messy, unproductive part of work doesn’t fit well in burndown charts, productivity metrics and token consumption. AI tools offer the illusion of a simple, abstracted solution. The popular narrative at the top is some version of “we can carry on as-is, don’t worry, everything will be amazing with AI.”

Like previous similar cycles of technology transformation and corporate profiteering, we will see a lot more damage before things improve. A lot of people will lose their jobs and suffer due to shortsighted, distant management. After a time, those companies will hire people back once they realise that all that human-to-human inefficiency isn’t easily replaced with a probabilistic, generative system. Or when, more carefully, quiet companies step up and fill the gaps. Eventually, the delusional returns will not materialise, and high-risk, high-return investors will pull back and look for something else to gamble on.

Also Read: Emotional intelligence makes AI training stick

The impact on knowledge workers will be terrible and cyclical. And while we are all distracted by this AI illusion, development funding cuts, war, environmental disasters, and global warming are taking away the futures of the children of the world. For the majority of workers, they will continue to do what they have always done, make the best of things in spite of clueless management. All of us will continue to make the best of the tools we are given, as we have done with ERPs, cloud computing and every iteration of the technology transformation eras. We aren’t going through an AI, cognitive boom. We are going through a social, political, and moral crisis. Developed economies have lost their way, established tech companies don’t know how to further juice their returns, and AI is the newest appealing distraction from real problems without easy solutions.

Humans won’t be replaced by robots or AI systems. It’s frustrating that we even need to debate this point. The debate alone shows us how far the social narrative has abstracted away from the reality of how our world works. The world is full of jagged edges that don’t fit in rapid-prototype-iteration cycles. AI tools are just another technology transformation trend. Like previous tech tools, they are amazing and offer lots of exciting new possibilities. And they are distracting us from the things that are harming humanity.

We cannot drink tokens, eat GPUs, live in data centres, or resolve wars with chatbots. The majority of the world is happening outside of the AI-obsessed global north. We used to chide each other about “the world happening outside our window,” yet now we are told to obsess over the world running on our phones and in our heads. Humans tire easily, and I expect we will soon tire of the obsession with superficial AI-enabled commerce.

Even a perfect AI system will be boring at a coffee table. We love to complain about each other and will go to great lengths to avoid interactions with each other. Yet we also love each other and find ways to care and support each other in so many ways. There is so much beauty and love in humanity. We don’t need AI to replace us if we just stop and look for a moment at what we have created, both the beauty and the horror.

We can imagine a better future for all of us with all kinds of technologies if we first invest in each other and the world outside of ourselves.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post The future is full of humans working with humans, AI systems and other technologies appeared first on e27.