- LLMs
- Chat assistants
- Gemini
AI-powered features
Building AI into real products: chat assistants, LLM-backed features and AI-assisted developer tooling. Sutra, the assistant on this site, is a live example of this work.
- Best for
- Products that need a chat assistant or an LLM-backed feature.
- Stack
- LLMs (Gemini, Groq), Next.js
What I build
Chat assistants
LLM-backed features
AI-assisted developer tooling
Details
Building AI into real products: chat assistants, LLM-backed features, and AI-assisted developer tooling. Sutra, the assistant on this site, is a live example of this work.
Proof from real projects
Platforms I've helped build in earlier roles, plus experiments on the same stack.
Experiments
- ReactJS
- NextJS
- Web App
- TypeScript
- LLM
- AI
- Generative AI
- Frontend
- +2 more
TopicTutor: AI Tutorial Generator That Spends Tokens Only Once
TopicTutor is an AI tutorial generator that turns any keyword into a level-based mini-course with runnable code and embedded videos, built from one cached model call.
Read the write-up- ReactJS
- Web App
- Tailwind CSS
- Vite
- shadcn/ui
- Radix UI
- lucide-react
- framer-motion
- +6 more
VisionCraft AI: Turning One Idea Into a Prompt You Can Actually Use
A dark, focused prompt studio for creators — you describe a video or image idea in a sentence or two, and VisionCraft AI hands back a polished, model-ready prompt without you ever needing to learn how to talk to an AI generator.
Read the write-up
What you can expect
Architecture first
A structure your team can extend, not a one-off build.
CI/CD and tests from day one
Every change is checked and built automatically before it ships.
Performance, SEO and accessibility
Server rendering, semantic HTML, metadata and Core Web Vitals are built in.
AI-assisted, human-reviewed
Coding agents for speed, with guardrails on the parts users actually feel.
Frequently asked questions
Can I see an example?
Yes. Sutra, the assistant on this site, answers questions from the site's own content, and TopicTutor turns any keyword into a mini-course from one cached model call.
Related writing
September 24, 2026
Where AI Agents Break on Frontend Work (and Where They Shine)
Agents are brilliant at frontend boilerplate and quietly bad at the parts users actually feel: design-system fidelity, accessibility, missing states and layout. Here's where they shine, where they break, and the guardrails I put around them.
Related services
React & Next.js web apps and dashboards
End-to-end web app development: data-heavy dashboards, internal tools and customer-facing products, built with React, Next.js, TypeScript and a proper CI/CD and testing setup.
Explore React & Next.js web apps and dashboardsReact Native & Flutter mobile apps
Cross-platform iOS and Android apps, mainly in React Native, so your team can reuse its React and TypeScript skills and logic from your web app. Flutter for smaller projects where it's the better fit.
Explore React Native & Flutter mobile appsFrontend architecture consulting & code review
An outside, senior review of your frontend: rendering strategy, data fetching, bundle size, Core Web Vitals and overall architecture, with a written report and concrete next steps.
Explore Frontend architecture consulting & code review
Let's talk about your project
Share what you're building, your timeline and your current stack. I'll reply with next steps and a personalised estimate.