Posts

Don’t rush that new MyKad. The card is ready. A lot of the apps aren’t.

Malaysia’s next-generation MyKad started rolling out nationwide on 17 September. It looks different, feels tougher, and comes with 53 security features — up from 23 — plus a polycarbonate body, NFC chip, hologram, transparent window, and a QR code on the back.

That’s the upgrade story. The practical story is messier.

Within days, the redesigned card was rejected during eKYC on Touch ’n Go eWallet, Ryt Bank, AEON Bank and GXBank. First-time MyDigital ID registration via the app also failed: the scan screen still expected the old layout, photo on the right, chip on the front. MyDigital ID says online support for the new card only starts on 1 October. Until then, first-time users need a physical kiosk.

JPN has confirmed the snag. eKYC providers are updating their systems. If your new card won’t scan, ask the bank or app for another verification method. Your old MyKad is still valid. You do not have to replace it.

Why the systems broke

The card didn’t just get more holograms. The layout changed enough to trip template-based scanners.

  • Photo moved from the right to the left (more like a driving licence)
  • Chip moved from the front to the back
  • Dual-interface chip: contact + NFC
  • No more embedded Touch ’n Go

If an app is still matching “photo right, chip left,” the new card looks “wrong” even when it is genuine. That is a coordination problem, not a security feature.

Malaysia has spent years talking about AI Nation 2030, MyDigital ID, and paperless government. Then the single most used identity document in the country changes — and the digital layer that depends on it is still catching up. A more secure card that cannot open a bank account or an e-wallet is only half a product.

What you should actually do

Keep your current MyKad if it is still in good condition. Existing cards work for government services and aid. There is no deadline to switch.

If you already have the new card:

  • Don’t assume every app will accept it this week
  • For MyDigital ID first-time signup, use a kiosk until 1 October
  • Ask banks and e-wallets for manual or alternative KYC
  • Get a standalone Touch ’n Go card if you used the old IC for tolls or transit — that function is gone

If you were about to queue at JPN “just because it’s new”: don’t. The Home Minister and JPN said the same thing before launch. This week proved why.

The new MyKad is a better physical credential. Until banks, wallets and MyDigital ID all read the same card on the same day, the smart move is patience — not a same-day reprint.

/by

Agentic AI: The Next Leap Beyond Chatbots

The AI conversation has shifted.

For the past two years, most of us interacted with AI through chat interfaces — ask questions, get an answer. Useful, but limited. Today, a more powerful paradigm is rapidly moving from research labs into real business systems: agentic AI.

What Exactly Is an Agentic AI Agent?

An agentic AI agent is an autonomous system that can perceive, reason, plan, and act toward a goal with minimal human intervention.

Unlike traditional generative AI, which waits for prompts, an agentic system:

  • Breaks complex goals into smaller tasks
  • Uses tools and APIs
  • Makes decisions based on intermediate results
  • Learns from outcomes and adjusts its approach
  • Collaborates with other agents when needed

In short, it doesn’t just respond — it gets things done.

How Agentic AI Works

Modern agentic systems typically combine several components:

  1. Reasoning engine, powered by large language models such as the latest Gemini or GPT-class models
  2. Memory, including short-term context and long-term knowledge
  3. Tool use, including search, code execution, databases, browsers, and internal systems
  4. Planning and reflection loops
  5. Multi-agent orchestration, where specialized agents hand off tasks to one another

The result is a system that can handle multi-step workflows that previously required constant human supervision.

Real-World Use Cases Emerging in 2026

  • Software engineering: Agents that diagnose bugs, write tests, open pull requests, and iterate based on CI feedback.
  • Customer support: Agents that resolve tickets end-to-end by accessing knowledge bases, CRM systems, and escalation paths.
  • Research and analysis: Agents that gather data from multiple sources, synthesize findings, and produce structured reports.
  • Operations and finance: Agents that monitor metrics, trigger alerts, reconcile data, and propose or execute corrective actions.
  • Sales and marketing: Agents that research prospects, personalize outreach, and schedule follow-ups.

Companies that have moved beyond simple copilots into true agentic systems are already seeing measurable gains in speed and cost efficiency.

The Challenges We Still Need to Solve

Agentic AI is powerful — and still imperfect. Key issues include:

  • Reliability and hallucination risk in long chains of actions
  • Security and access control when agents are given real tools and permissions
  • Cost management, as agents can consume significantly more tokens than single-turn interactions
  • Evaluation and observability — measuring whether an agent achieved the goal
  • Human oversight design — determining when an agent should stop and ask for approval

The organizations succeeding with agents treat them as systems that require engineering discipline, not magic.

Looking Ahead

We are moving from “AI that answers questions” to “AI that completes work.”

The winners in the next 18–24 months will not be those who simply adopt the latest model, but those who thoughtfully design agent workflows, guardrails, and feedback loops. Agentic AI rewards clarity of goals, strong tool integration, and continuous evaluation.

The question is no longer whether agents will become part of how work gets done. The question is how quickly, and safely, each organization will integrate them.

/by