SG Code Campus

If a feature cannot be implemented cleanly in code on the first pass, prompting more is unlikely to fix it. The issue is usually not the model. It is the specification.

That was the clearest pattern we took away from Google Cloud Next 2026. As a Google Authorised Training Partner, we spent the week looking at how teams are actually shipping with these models, and the ones moving fastest were not prompting harder. They were writing better specs and treating ex*****on as an engineering problem.

The Gemini CLI reflects this. So does FairPrice Group's Store of Tomorrow, which was one of the more grounding local examples on the floor.

The constraint is no longer access to capable models. It is whether your team can write the spec.

Thanks to our team, Trevor (COO) and Huiliang (CDO) for representing on the ground.

Is your team moving towards a spec-first approach, or still relying on prompt iteration?

Our fourth Agentic Builders Collective (ABC) meetup at Lorong AI was a masterclass in the technical realities of agentic AI. Huge thanks to Lorong AI for hosting us at this government-backed hub to explore the practical breakthroughs defining the frontier of AI engineering.

Keypoints from our speakers:

🔄 Boon Kgim analyzed "context rot" and the 40% wall, where attention decay causes agents to fail. He proposed Workflow Engineering 2.0, using Generator + Evaluator loops to achieve senior-level quality. His Thin Orchestrator coordinates specialized workers via file artifacts to protect context budgets.

📁 Gaurav Keerthi argued that AI must be a "resident" of the workflow, not just a visitor. Instead of fiddling with general tools, he built Workforce-OS, a file-based system tailored to his team’s specific stack and needs. Using markdown and git as a Kanban board, he proved that highly customized, bespoke AI is both possible and perfectly auditable.

🛡️ Gaurav Manek presented a masterclass on Infrastructure-based Safety, framing errors as predictable failures. Using the STRIDE model, his framework focuses on pillars like minimizing blast radius and ensuring reversibility. He demonstrated an MCP Gateway that mediates access via rate limits to create a hardened control plane.

⚖️ Darryl Wong explored using Codex to build defensible financial platforms. He detailed a 9-stage state machine for bookkeeping where every step mandates strict validation. To ensure integrity, he used a Mathematical Validation Gate and a Readiness Gate to separate planning from implementation.

These insights are already shaping our work. Following our recent Farrer Park Hospital delivery, we are applying these patterns to new optimization projects and experimenting with hybrid workflows: using Claude for planning and Codex for building.

Shout out to the team:

💡 Chris Pecaut for emceeing and keeping the insights flowing.

💡 Jensen Loke for his relentless community drive.

💡 YJ Soon for ensuring our ecosystem thrives.

ABC is at the precipice of a major shift. this level of technical sharing defines the leaders of the AI transition, so what are you building at this frontier? 💻✨

Four crews. Three Days. Two Events. One summit. Zero chill. ⚡

SGCC, THCC, JPCC and MY Code Campus came together at AWS Summit to learn, chill and connect and as always, it lived up to the hype. We kicked things off the night before at Cook and Brew, because no Code Campus event starts without a meal first 🍻

Here's what stood out 👇

🤝 MEETING THE Amazon Web Services TEAM & PARTNERS
Every conversation with our AWS friends made the day. Thank you to the AMs of the Public and Commercial Sector for swinging by despite your back to back meetings. KM, Pabitra and Woan appreciate you checking in on us multiple times. Joey, thank you for the ever-bubbly intros to your ASEAN colleagues 💛

🚀 SHOWCASING OUR HACKATHON PLATFORM
We officially debuted our high engagement hackathon platform built to help senior management bridge AI theory and real world implementation. It's for leaders who want to drive digital transformation by doing, not just observing. First public showing, but we've already run it with Cognizant and UTHM 💪

📚 ACCESSIBLE TRAINING & CERTIFICATION
So many of you stopped by asking how to stack badges in the quickest, most cost effective way. AWS certs are clearly in demand the conversation has moved from "should I do it?" to "how fast can I get it?" 👀
Through our partnership with SMU Academy, we're making high level training accessible from foundational cloud skills to specialized AI and ML.

Meeting everyone was the highlight. Thank you for the incredible energy 💜

🎁 Joined our raffle? Winners drop Wednesday, 13 May 2026. Stay tuned!

We're at the edge of a major shift in how we build and scale.
So what are you building at this new frontier? 💻✨

How do educators shift from AI curiosity to building functional tools in one afternoon? 🚀

On April 10th, SG Code Campus facilitated an intensive AI Learning Journey for over 70 Singapore Polytechnic staff to bridge the gap into the "Agentic" era. We moved beyond simple chatbots to explore AI as a proactive partner in the classroom.

Here are the highlights from the session:

🛠️ The Power of Vibe Coding
Watching educators with zero coding background deploy 15 functional web apps in under 20 minutes was incredible. By focusing on natural-language intent rather than code, they built tools like a Student Rating System and a Transformation Matrix Visualiser to solve real academic bottlenecks.

📚 NotebookLM: The Standout Tool
NotebookLM resonated quickly, cited as a standout by 78% of the room. Educators saw its potential to save hours of administrative work by summarizing research and instantly generating quizzes or mind maps.

🧠 Architecting the Reasoning Loop
We focused on "Architecting Reasoning Loops," moving past simple prompts to systems that evaluate and self-correct. For high-stakes tasks like internship grading, we prioritized the professional guardrails needed to keep tools pedagogically sound.

💡Bridging the Gap to Action
With 65% of the group planning immediate classroom use, the workshop was a turning point. The focus shifted from experimentation to practical AI engineering.

A huge thank you to Felix Goh (Gemini Adoption Lead, Google) for the close collaboration. We also thank Edna Chan (Director) and Lee Hwee Theng (Deputy Director) of the School of Mathematical Sciences & Analytics at Singapore Polytechnic for their trust. This commitment to innovation is what makes these breakthroughs possible.

We are proud to support these educators as they redefine the classroom. We can’t wait to see these prototypes become daily essentials. 💻✨

Our third meetup was a major milestone for the agentic coding community! 🚀

Big thanks to Felix Goh for hosting us at Google HQ. Great food, beer, and hospitality provided the perfect backdrop for the community’s energy and technical depth.

To reflect the evolving landscape, we have officially rebranded from Claude Code Club to the Agentic Builders Collective (ABC). This shift is about inclusivity and staying at the forefront of agentic coding, regardless of the tools used.

Here are the key takeaways from our speakers:

1️⃣ Hui Liang gave an expert’s look at Google’s Antigravity, breaking down production use cases and pitfalls for power users.

2️⃣ YJ Soon explored the landscape beyond Claude and Codex, sharing how to navigate today’s explosion of agentic options efficiently.

3️⃣ Ian and Natasha shared their journey building an Admission and Discharge platform for Farrer Park Hospital. They discussed the hybrid AI system that effectively combines the reliability of deterministic rules for fixed costs with the adaptability of LLM few-shot learning for variable consumables, creating a robust and highly accurate cost estimation system.

4️⃣ Logan Goh challenged us to drop prompt deference. He shared how assertive prompting, high-precision personas, and mechanical enforcement via hooks solve "Degeneration-of-Thought" to build reliable AI agents.

The night proved we are moving from experimentation to true AI engineering. Success now relies on architecting the validation loops and deterministic cores needed to harness LLMs effectively in the real world.

Thanks to the team for the seamless ex*****on:
💡 Chris Pecaut for masterfully emceeing and keeping the insights flowing.
💡 Jensen Loke for his relentless drive in building this momentum.
💡 YJ Soon for the steady behind-the-scenes support ensuring everything runs perfectly.

ABC is just getting started. This level of technical sharing defines the leaders in this AI transition.

What are you building at the frontier of agentic coding? 💻✨

A decade of excellence at Farrer Park Hospital! 🏥✨

Earlier this month, we had the honor of attending the Farrer Park Hospital 10th Anniversary Gala. While the night was a beautiful reflection on the hospital’s journey, from its vital role during COVID-19 to its growth as a premier medical center, the true focus was on the future of healthcare.

The highlight of the evening was hearing Farrer Park Company Board Chairman Maurice Choo identify AI as a top priority for the hospital. He shared a sincere thank you to the five core partners working in lockstep to bring Farrer Park Hospital into the AI era: SG Code Campus, AWS, Cloud Kinetics, SMU Academy, and Apple.

Achieving this vision requires more than just high-level strategy; it takes deep alignment across the entire ecosystem. It was a privilege to share a table with the leaders driving this mission forward. Our leadership team, Ian (CEO), Trevor (COO), and Hui Liang (CDO), were joined by Natasha Waliany, Head of AI, and Tan Pek Yee, Head of DIG at Farrer Park Hospital. We were also joined by Joe Francis, President of Commercial, and Wilson Chew, MD for Singapore at Cloud Kinetics, alongside Binny Peh, Director for Healthcare at AWS, and Cara Lee from AWS Public Sector.

Having all four organizations in one room underscores the weight behind this partnership. We aren’t just discussing "digital transformation" in theory; we are architecting an ecosystem that brings production-ready AI to healthcare operations.

This collaboration is already delivering tangible results through the rollout of an AI-enhanced Admissions System. By integrating an in-house model designed to provide patients with clearer financial counseling, we have worked together to reduce administrative friction and improve transparency. Beyond the deployment, our team spent time on-site to train admissions officers, ensuring the frontline feels confident using these new tools to enhance the patient journey at Farrer Park Hospital.

Huge congratulations to the entire Farrer Park Hospital team on ten years of redefining the medical scene. We are proud to be one of the five partners helping you build the future of healthcare.

Bridging the Gap Between AI Ambition and Real-World Care 🏥✨

We are proud to share the successful rollout of a new Admissions System for our client, Farrer Park Hospital, in close collaboration with our infrastructure partner, Cloud Kinetics.

In healthcare, digital transformation is about more than just efficiency. It is about the people behind the process. By streamlining the patient admissions workflow, we have worked to reduce administrative friction so that the focus remains where it belongs: on the patient.

What’s new in this deployment?

AI-Enhanced Clarity: We have integrated an in-house AI model specifically designed to provide patients with more accurate and transparent financial counselling. 🤖📊

Support for the Frontline: Technology is only as effective as the team using it. We were on-site to train admissions officers and provided live support throughout the transition to ensure everyone felt confident and capable. 🤝

A Stronger Ecosystem: This project highlights what happens when specialized partners like SG Code Campus, an AWS Advanced Tier Training and Consulting Partner, and systems integrators like Cloud Kinetics, an AWS Premier Tier Services Partner, work as a single ecosystem to deliver production-ready AI.

A big thank you to the Farrer Park Hospital team for their trust and to Cloud Kinetics for the seamless infrastructure. We are excited to see this system in action, making a tangible difference in the patient journey! 🚀

What does it actually take for universities to help organisations succeed in AI transformation?

More than just a pilot project or a slide deck. It requires a practical model where academia, training partners, and industry implementers work together as a single ecosystem.

Last week, SG Code Campus (SGCC) and SMU Academy (SMUA) had the privilege of hosting the executive delegation from the University of the Thai Chamber of Commerce (UTCC) for a deep dive into sustainable digital transformation.

Through this collaboration, a practical model is emerging. We are moving organisations from AI ambition to real deployment. Here are the core themes from the session:

1️⃣ Bridging the Fluency Gap
Digital transformation is ultimately about people. Institutions like SMUA help leaders build the technical fluency needed to guide AI initiatives. This ensures they maintain control over scope, cost, and accountability.

2️⃣ The Power of the Ecosystem
We explored how a partner ecosystem creates a practical pathway for adopting AI. Specialised partners like SGCC support capability building, while system integrators like Cloud Kinetics provide the managed services and infrastructure required for production.

3️⃣ Production-Ready AI: A Client Success Story
Natasha, Head of AI at our client Farrer Park Hospital, shared how LLM-based systems are being used in production. By supporting financial counselling and patient admission workflows, the system streamlined processes that previously required significant manual coordination. It provided a practical example of how our AI implementations improve operational efficiency. 🏥✨

4️⃣ Architecting with Intent
The game has changed from writing lines of code to architecting outcomes. Ian, our CEO, and Eric, our Head of Technology, demonstrated the shift toward Agentic Coding using Claude Code, where a concept can be translated into a functional prototype within a single afternoon. 💻

A sincere thank you to our hosts at SMUA, especially Jack Lim and Harry Lee, for facilitating such a thoughtful environment for exchange. The engagement from the UTCC team was energizing, and we look forward to continuing the conversation in Bangkok! 🇹🇭✈️

The energy at the second Claude Code Club meetup three weeks back proved one thing: the future of software development isn't just about better models, it’s about the community figuring out how to use them together.

SG Code Campus was proud to sponsor the venue with our partners at the Lifelong Learning Institute to officially welcome the new era of environmental fluidity and agentic workflows. We hope everyone enjoyed the sumptuous buffet dinner on us as well, because let's face it, there’s nothing like good food to fuel a deep dive into the world of AI!

Here are three core takeaways from the session:

1️⃣ The Death of the "Traditional" Workspace:
Gijs Verheijke (Carousell) shared the journey of building markjason.sh. The highlight? A story about coding from a bathtub using Clawdbot. While humorous, the implication is profound: AI tools have lowered the friction of creation so significantly that the "where" and "when" of engineering are being completely redefined.

2️⃣ Is Language Our New Bottleneck?:
Jeff Stryer posed a provocative question: Is human language, both programming and otherwise, actually limiting what AI can produce? We had a deep dive into whether our current way of "talking" to models is holding back the next leap in code generation. As we move toward more complex systems, the art of the prompt is evolving into the science of intent.

3️⃣ Agentic AI as the Future of EdTech:
Coming from an education background, Pallani’s demo of MCQ.sg really hit home for us. Seeing agentic AI build practical tools for the next generation of learners makes it clear: AI agents aren't just a trend, they are the new infrastructure for how we will teach and learn.

A Note of Gratitude to:
💡 Jensen Loke for his relentless drive in keeping this momentum going.
💡 Chris Pecaut for masterfully moderating the chaos and keeping the insights flowing.
💡 YJ Soon for the seamless ex*****on behind the scenes.

The most inspiring part of that night? Seeing everyone sharing screens and asking questions. That willingness to experiment in public is exactly what will define the leaders in this AI transition.

So what are you building with Claude?

It’s been a month since we wrapped up an incredible session at Google’s Cloud Technical Series OnBoard Edition! 🚀

The energy from the developer community was a clear signal: the era of multi-modal AI and intelligent agents is here. As Google Authorized Training Partners, SG Code Campus was proud to have our resident Google Authorized Trainers, Hui Liang and Natasha, guiding everyone through the practical side of these frontier models.

Here’s a recap of the hands-on magic we covered:

🧠 Lab 1 - Gemini Pro Unlocked: We went under the hood with Google’s most versatile model. Participants mastered multi-modal reasoning, using the Gemini API and Gen AI SDK in Jupyter environments to process text, code, and images simultaneously.

💻 Lab 2 - Building with ADK: We moved into the "agentic" era with the Agent Development Kit. From building an agent with search tools to testing via CLI and programmatic scripts, we saw firsthand how ADK streamlines the development of intelligent assistants.

A big thank you to Google Cloud for having us, and to every developer who joined us to level up their Gen AI game. We can’t wait to see what you build next! 🛠️✨