AI Weekly Malaysia

Andrew Ambrosino, OpenAI Codex lead, explains how AI makes software cheaper and faster to build, shifting focus from coding to product taste and user experience. The Codex desktop app lets non-developers create working apps, lowering barriers for rapid prototyping. This trend rewards strong product intuition over traditional engineering scale.

Why: Malaysian startups and builders can now prototype and deploy products at a fraction of the cost and time, emphasizing local market insight and design over large engineering teams. It democratizes software creation, enabling more founders to test ideas quickly.

WhyQ, a Malaysian food delivery startup, spent a decade pivoting from hawker-to-office delivery to residential delivery during COVID, and finally found a sustainable model in corporate dining. The article outlines how the company learned that scaling too fast burned cash, and a focused B2B approach now delivers better unit economics.

Why: Startup founders can learn from WhyQ's pivot: rapid B2C scaling without solid unit economics is a trap, and a lean B2B model (corporate meal plans) can unlock profitability in a competitive food delivery market. The lesson is directly applicable to SaaS and marketplace founders in Southeast Asia.

This week's Lenny's Newsletter covers two main topics: a review of the GLM-5.2 open-source LLM from China, which is competitive with top models on coding and reasoning benchmarks, and a deep dive into how Gusto used Claude Code to rapidly build a new product line, highlighting practical AI agent workflows for engineering teams.

Why: For developers and AI/ML learners, the GLM-5.2 review offers a cost-effective, open-source alternative to proprietary models. For startup founders and engineering leaders, the Gusto case study provides a real-world blueprint for using AI coding agents (like Claude Code) to accelerate product development, which is directly applicable to building in Southeast Asia's cost-sensitive market.

Gusto CTO Eddie Kim shares how a 5-person team shipped a new AI product line in 10 weeks using Claude Code, a permanent Zoom call, and no Figma, Jira, or traditional docs. They relied on AI-generated code, rapid prototyping, and blurred engineering/product roles.

Why: Demonstrates a radical, low-overhead approach to building AI products fast—relevant for Malaysian startups and developers looking to compete globally with lean teams and minimal tooling costs.

The Ministry of Digital has launched a national AI transformation initiative to accelerate AI adoption across Malaysia's public and private sectors, likely involving policy frameworks, infrastructure, and talent programs.

Why: This could unlock government grants, sandboxes, and contracts for local AI builders, while shaping the regulatory environment for AI development and deployment in Malaysia.

Asian AI startups are releasing models with capabilities comparable to Anthropic's upcoming 'Mythos' line, capitalizing on prolonged US export restrictions that limit access to frontier American models. The shift threatens to permanently redirect Southeast Asian and broader Asian demand toward domestic and regional providers. US labs risk losing one of the fastest-growing AI markets if the export ban continues.

Why: Malaysian builders and founders may soon have credible, locally-hosted frontier-tier alternatives that avoid US export controls, lower latency, and potentially offer better pricing in MYR-friendly terms. Choosing an AI stack now means weighing whether to bet on US frontier models or hedge with regional providers that are rapidly closing the capability gap.

Ornith-1.0 is a new open-weight coding model from DeepReinforce, built on Gemma 4 and Qwen 3.5, with sizes up to 397B parameters, achieving top open-source performance on coding benchmarks. It excels at agentic coding with multiple tool calls and runs locally via LM Studio, making it accessible for offline use. Early tests show strong code search and manipulation capabilities.

Why: It provides a free, open-source alternative to commercial coding agents, enabling local, privacy-respecting, and customizable coding assistance for developers and AI agent users.

Hasan.VC concluded its Fund I accelerator with a Demo Day in Bandung, showcasing 20 startups from Cohort 004. Over four cohorts, the programme supported 120 startups and nearly 500 founders across 10 countries, using a people-powered halal venture capital model. The event included live pitches, panels on startup fundability, and a closed-door investment session.

Why: It highlights a growing regional accelerator model with halal VC, offering Malaysian founders insights into what makes early-stage startups fundable and potential funding pathways.

A new report warns that memory (DRAM/NAND) prices will keep rising until 2028 due to a persistent supply-demand gap, with only 60% of demand met by 2027. This extends the ongoing shortage driven by AI, data centers, and device production, directly impacting hardware and cloud costs.

Why: Higher memory costs will inflate cloud bills, server expenses, and device prices for Malaysian startups and developers. Local data center builds and AI/ML projects may face budget overruns, forcing trade-offs in infrastructure planning.

Digital Penang and OSK Ventures have signed a one-year MoU to improve access to venture debt and equity financing for AI, hardtech, and deeptech startups in Penang, targeting the capital gap faced by companies with long development cycles and high capital needs.

Why: This directly addresses a critical bottleneck for Malaysian deep tech founders—access to growth-stage capital. It signals institutional support for startups that don't fit traditional financing models, potentially accelerating commercialization and scaling for AI and hardware ventures in the northern region.

Simon Willison amplifies Jon Udell's call to reframe 'human in the loop' as 'agents in our loop,' arguing developers should invite AI agents into existing, reviewable workflows rather than ceding authority to black-box agentic processes that produce unreviewable outputs.

Why: For developers and vibe coders building with AI agents, this directly shapes how you design agentic pipelines—keeping code review, testing, and ownership human-first prevents unmaintainable messes and keeps you in control.

HP Inc. expands its 'Frontier' partnership with OpenAI to integrate advanced AI models into customer support, internal software development, and enterprise workflows. This signals a major PC manufacturer embedding AI deeply into its product lifecycle and operations, not just consumer features.

Why: For builders, when a hardware giant like HP treats AI as core infrastructure for software development and operations, it validates that AI-assisted engineering and enterprise tooling are becoming table stakes. It also hints at future enterprise demand for AI-optimized hardware and locally run models on PCs, affecting where your apps might run.

South Korea announced a massive US$576 billion investment plan with Samsung and SK Hynix to expand AI chip and semiconductor production, aiming to dominate the global AI hardware supply chain amid surging demand and ongoing memory shortages.

Why: Malaysia is a major semiconductor packaging and testing hub; this signals intensified regional competition for talent, investment, and supply chain positioning. Local tech firms and data center operators may face tighter chip supply and pricing pressure.

Simon Willison's HTML Table Extractor is a paste-conversion tool that extracts tables from copied rich text and converts them to HTML, Markdown, CSV, TSV, or JSON. It's particularly handy for quickly grabbing structured data from web pages like Wikipedia lists. A related update improves his Rich text to markdown tool with table support.

Why: Saves developers time when scraping or repurposing tabular data from websites, which is common when building dashboards, importing data into databases, or preprocessing for AI/ML. For Malaysian builders, it’s a practical utility for extracting data from local government portals, research reports, or any table-heavy site.

Chamath Palihapitiya raised a $135M Series A for his AI coding startup and stepped in as CEO, signaling continued heavy VC appetite for AI developer tools despite a crowded market.

Why: Validates that AI coding assistants are still a hot funding category. For builders, it means more competition and better tools, but also potential vendor lock-in and pricing shifts as well-funded players scale.

Arena, the widely-used AI model leaderboard, has become a $100M business just months after launching commercial services. The platform helps developers and enterprises benchmark and compare AI models. Its rapid growth signals strong demand for AI evaluation tools.

Why: It shows that model benchmarking is becoming a key business, influencing how developers and companies choose and pay for AI models. The leaderboard’s ratings directly impact purchasing decisions and development workflows.

The JAKIM Sabah website was defaced by attackers claiming affiliation with Anonymous, who allege they stole admin emails. The site was taken offline, and JAKIM has not yet publicly addressed the breach.

Why: Highlights critical cybersecurity vulnerabilities in government digital services, underscoring the need for developers to enforce robust security practices and for startups to consider the reliability of public-sector partners.

NTT Data Malaysia's MD Henrick Choo argues that Malaysian enterprises risk being outcompeted if they don't invest in AI, even if ROI is uncertain. He notes Malaysian firms lag Singapore by ~2 years in AI adoption, but spending continues, with NTT targeting 20% growth. He also stresses Malaysia's need to develop higher-quality AI and cloud engineers to capture global opportunities.

Why: Validates that Malaysian enterprise AI spending is real but cautious. For builders, it signals growing demand for AI integration services and a critical talent gap in AI/cloud engineering that local developers can fill.

OpenAI has previewed GPT-5.6 Sol, positioning it as a next-generation model with notable improvements in coding, scientific reasoning, and cybersecurity tasks, alongside an upgraded safety stack. As a preview release, availability and pricing details will shape how quickly developers can integrate it into production workflows.

Why: Stronger coding and reasoning capabilities directly affect developer productivity, agent reliability, and the kinds of SaaS features founders can ship without hiring specialists. The enhanced safety stack also matters for Malaysian teams building customer-facing products where compliance and responsible-AI scrutiny are increasing.

OpenAI reports that median internal Codex output tokens grew 56x in Research, 32x in Customer Support, 27x in Engineering, and 13x in Legal since November 2025. The growth is steepest in non-engineering teams, suggesting Codex is being adopted for scripting, automation, and document tasks well beyond traditional software development. The data is a self-reported internal benchmark, so it reflects usage volume rather than validated productivity gains.

Why: For Malaysian devs and builders, this is a concrete signal that AI coding agents are no longer just for software engineers — support, legal, and research workflows are now measurable Codex use cases. If you sell to or build for non-technical teams, the 'AI for non-coders' wedge is already being validated by the lab shipping the tool.

Hugging Face now lets you spin up a vLLM inference server on HF Jobs with a single CLI command, removing much of the boilerplate around provisioning GPUs and configuring the vLLM runtime. The post walks through launching an OpenAI-compatible endpoint, pointing an existing client at it, and tearing the job down when finished.

Why: For the community, this lowers the cost of experimenting with self-hosted open-weight models. Instead of renting a GPU, installing CUDA drivers, and wiring up vLLM manually, you can go from zero to a working inference endpoint in minutes, which is ideal for demos, coursework, or short-lived benchmarking sessions.

DiSCoFormer introduces a single transformer model that jointly learns both the probability density function and the score function (gradient of log-density) across multiple distributions. This unified approach enables tasks like sampling, density evaluation, and out-of-distribution detection without needing separate models.

Why: For AI/ML practitioners, a single model that handles both density estimation and score matching can streamline generative modeling pipelines, reduce maintenance overhead, and potentially improve sample quality and evaluation speed.

Google UK's Economic Impact Report argues that widespread AI adoption could boost UK productivity by over £400 billion by 2030, but requires a national push in skills training, especially for SMEs and public services. The report highlights practical tools like AI-powered assistants and free training programs to democratize access.

Why: The report's emphasis on practical, accessible AI tools and free training mirrors the upskilling needs of Malaysia's own SME-heavy economy. It provides a blueprint for how local developers and founders can leverage similar 'AI trailblazer' initiatives to drive productivity without needing deep ML expertise.

A Google AI expert explains the concept of a full-stack AI approach, covering integrated hardware and software layers from TPUs to developer tools. This deep integration optimizes performance, scalability, and efficiency for AI workloads.

Why: Understanding the full AI stack helps builders choose platforms wisely and debug performance issues. It demystifies cloud AI services and guides decisions on whether to use managed solutions or build custom stacks.

Omen AI raised a $31M Series A to build sensors and AI models that monitor liquid coolant quality in data centers, detecting bacterial growth and corrosion before they damage GPU clusters. The system targets the growing liquid-cooled infrastructure powering AI workloads.

Why: As Malaysia and neighbors like Johor rapidly build hyperscale data centers for AI and cloud workloads, liquid cooling is becoming standard. A practical operations tool that prevents downtime and hardware degradation directly impacts cloud cost and compute reliability for builders and startups relying on these facilities.