Tech Talk

PDPA: How A ChatGPT Prompt Could Cost Directors RM250,000

A single wrongly sent email or an unauthorised ChatGPT upload can now trigger a statutory 72-hour reporting deadline under Malaysia's updated PDPA. Ong Johnson and Lo Khai Yi of Halim Hong & Quek join Tech Talk to explain why "simple negligence" is legally indistinguishable from a hacker attack, the RM250,000 fines facing directors, and why a fire drill for your data is no longer optional.Tune In To Learn More About:The "Output" Doctrine: Why the law disregards your intent, whether it's a sophisticated ransomware hit or a misplaced paper form, the legal breach is defined solely by the effect on personal data.The 72-Hour Clock: The strict timeline for assessing "significant harm" and why internal delays for board meetings or procurement can lead to a "love letter" (investigation) from the Personal Data Protection Department.The LLM Leak: The hidden risk of employees using unpaid, consumer-grade AI chatbots to process client lists, effectively training global models on your private data.Director Accountability: How failing to notify regulators doesn't just hurt the company; it exposes individual directors to personal liability and potential two-year imprisonment.The Multi-Jurisdictional Logistical Nightmare: Managing live updates and conflicting timelines across APAC, Europe, and the US during a global breach.Simulations vs. Policies: Why everyone "has a plan until they get punched in the face", the critical need for annual breach simulations over static manuals.Data Processing Agreements (DPA): The "10-out-of-10" requirement to mandate that third-party suppliers return or destroy data once a relationship ends.Tech Talk is brought to you by Maxis Business. Building Tomorrow’s Business Today.See omnystudio.com/listener for privacy information.

MSP359. The Social Cost Of Scarcity

Technology can produce more goods, more services and more value. Heck, froyo for all. But who is going to control this endless abundance? Our tech leaders envision a world where AI and automation make work obsolete and there’s wealth for all. Is this a social vision for a better future or just power and influence in a shinier wrapper?Matt and Roshan unpack the dream, the delusion, and the social math behind these notions of post-work abundance.See omnystudio.com/listener for privacy information.

Stop Slapping AI On Broken CX Processes

Why do we have more AI than ever, yet customer service feels increasingly "broken"? Raymond Davadass, CEO of Daythree and Chair of GBS Malaysia, reveals the "fragmentation trap" sabotaging large-scale operations. From why a misplaced comma can paralyse a bot to the 12-week design rule for AI deployment, we unpack why the industry is ditching "labour arbitrage" for a high-stakes play on data intelligence and human empathy.Tune In To Find Out:The Orchestration Gap: Why having a "beautiful faucet" of technology is useless if your internal systems aren't talking to each other (and the airport chatbot story that proves it).The "Comma" Crisis: Why AI isn't a magic wand and how an uncleaned knowledge base leads to "hallucinations" and customer frustration.Human-at-the-Right-Time: The specific moment in a customer journey (like lost luggage) where technology must step back and let human empathy take the lead to save the brand.The New Litmus Test: Why "Headcount" is a dead metric and why "Revenue per Employee" is the new standard for the next decade of business process management.Tech Talk is brought to you by Maxis Business. Building Tomorrow's Business Today.See omnystudio.com/listener for privacy information.

$100K AI Bills? 4 Drivers of Token Bill Shock

Is your company’s AI strategy a ticking financial time bomb? While consumer subscriptions offer "subsidised intelligence," the transition to enterprise agents like OpenClaw is exposing businesses to "AI bill shocks" and context window bloat. From the hidden mechanics of token-based billing to the capital expenditure of moving models on-premise, Daren Tan of ALPHV Technologies joins Tech Talk to unpack how to scale AI without breaking the bank.Learn More About:Token Billing Mechanics: AI costs are driven by "tokens", broken-down text fragments, where input data (prompts/PDFs) and output responses are billed at different rates depending on the model's reasoning complexity.The "Context Window" Bloat: As chat histories grow, LLMs re-process the entire conversation for every new prompt, exponentially increasing costs. Daren recommends starting fresh threads to prevent "context rot" and wasted spend.Four Pillars of Bill Shock: Organisations face financial risk from 24/7 agentic loops, "agentic checking" (AI verifying AI), and the illusion that enterprise API costs mirror $20 consumer subscriptions.Model Routing & Caching: Strategic cost-saving involves "model routing", sending simple tasks to cheaper "Nano" models, and utilising prompt caching to receive up to 90% discounts on recurring queries.The On-Premise Pivot: When cloud costs become commercially unviable, shifting to local hardware (like Nvidia DGX systems) offers fixed costs and mandatory data sovereignty for regulated industries, albeit with significant upfront Capex.The Governance Gap: Deploying autonomous agents like OpenClaw without strict oversight can grant AI full access to file systems, creating critical security vulnerabilities and credential risks.Tech Talk is brought to you by Maxis Business. Building Tomorrow's Business Today.See omnystudio.com/listener for privacy information.

MSP358. Back To Basic: The Universal Town Square

Free buses, childcare, broadband and housing; who decides what makes a universal basic right? This week, Matt and Roshan discuss Universal Basic Services and begin a two-part conversation on social resilience, post scarcity abundance, and the techno utopians battling to shape and profit from the future.See omnystudio.com/listener for privacy information.