Our Tech Stack

Next.js and React are the foundation of almost everything we build for the web. Next.js gives us server-side rendering, static generation, and API routes in a single framework, which means faster load times, better SEO, and a codebase that stays coherent as it grows. React's component model keeps the UI maintainable across complex, multi-page applications. For mobile, we build in React Native and Expo. Expo accelerates the development cycle significantly, handling the build infrastructure so we can focus on the product. A single codebase runs on both iOS and Android, with native performance where it counts.

Sanity is our headless CMS of choice. It separates content from presentation, which means the people running the business can update, publish, and manage content without touching the codebase. Sanity Studio is fully customisable, so the editing experience is built around the actual content model rather than forcing a generic interface on a specific problem. It also integrates cleanly with Next.js, which keeps the stack coherent.

Firebase is our default for real-time data, authentication, and serverless backend functions. It scales without requiring infrastructure management, handles authentication reliably, and connects well with the rest of the stack. For projects that need a relational data model or more SQL-native querying, Supabase is our PostgreSQL-based alternative, open source and increasingly mature. For more complex backend requirements, we evaluate the right tool for the brief. The guiding principle is always: what is the simplest thing that will handle the scale this product needs, without introducing unnecessary complexity.

Google Cloud Platform is our primary cloud environment, covering compute, storage, managed services, and AI infrastructure. For projects where AWS is already the client's standard or where a specific AWS service is the right fit, we are comfortable working within that ecosystem too.

Vercel is our preferred hosting platform for Next.js applications. It is built for the framework, deploys on every push, and handles edge functions, serverless, and CDN out of the box. For more traditional hosting needs, or where a client already has an existing relationship with a provider, we work with Hostinger and GCP directly.

This is where the stack gets interesting. For most AI-augmented features, we work with Anthropic's Claude API, which powers conversational interfaces, document intelligence, content generation, and agentic workflows. We also integrate with OpenAI and other model providers depending on the specific capability required. Mistral AI is a growing part of our toolkit, particularly for Singapore and Southeast Asian public sector and government engagements. Mistral is listed in IMDA's Innovative Tech Companies Directory as an accredited AI company, making it a trusted choice for projects where regulatory alignment with Singapore's national AI strategy matters. Its open-weight models also offer strong options where data sovereignty, on-premise deployment, or cost efficiency are priorities. For enterprise-scale AI deployments and projects requiring more control over the model layer, we work with Google Vertex AI and AWS Bedrock, both of which support open-weight models alongside proprietary ones. The model choice is always driven by the problem, not by preference.

Stripe for payments, subscriptions, and billing. Resend for transactional email. Twilio for SMS and messaging workflows. Beehiiv for newsletters and content distribution.

We do not impose a stack. We start with the problem and the constraints, which include the client's existing infrastructure, the team that will maintain the product after launch, the expected scale, and the budget. The tools listed here are the ones we know well enough to move fast with and trust under pressure. If a project calls for something outside this list, we will say so honestly rather than retrofitting the wrong tool.