Products, tools, and platforms I've built — some open source, some shipped inside an agency, all designed to do one specific job well.
A lightweight (~30 MB) native code editor with first-class AI. BYOK Anthropic, OpenAI, Ollama, or sign in with the Claude Code CLI to use your existing Claude Pro/Max subscription — no extra API key. Multi-workspace, pop-out terminals, zero telemetry, Tauri 2 + Rust.
All-in-one business platform replacing the Xero + HubSpot + Calendly + Mailchimp stack with one integrated app. GST-compliant invoicing with auto reminders, 24/7 online bookings, full accounting, CRM with activity timelines, email/SMS marketing, and an AI Companion for payment chasing and review requests. Free plan, paid from $24/month, native iOS app live.
Real-time results dashboard built during Nepal's 2082 federal election — FPTP, PR, province-level breakdowns, party color mapping, an interactive Leaflet map, and a CORS-safe data layer that pulled from both ECN (via NepalVotes) and Hamro Patro. Now archived: the eleven API responses are snapshotted locally so the dashboard renders without polling either provider.
Educational platform and AI-driven podcast (co-hosted with J and Mini) on the AI-SEO intersection. Frameworks, training, and research adopted by multiple agencies.
A curated, opinionated list of the SEO tools, browser extensions, and APIs I actually use day to day — so other operators don't have to rebuild their stack from scratch.
Proprietary AI tooling and automation platform built for Supple's enterprise SEO operations. Reduced manual operational overhead by ~40% and standardised QA across 100+ accounts.
A short, honest list. Builders pretending nothing fails are usually the ones to be skeptical of.
What: a real-time tracker for Australian skilled migration invitation rounds. Why it stopped: the data flow from Home Affairs became too inconsistent to keep current without manual upkeep. Lesson: upstream data reliability decides whether a tool can exist, not user demand.
What: a tiny fetch_weather.php on the homepage. Why it stopped: third-party API key churn and rate limits made it brittle. Lesson: every external dependency is a future maintenance bill — only take the bill if the value justifies it.
What: a prior multi-purpose Node experiment that lived under /mynodeapp/. Why it stopped: scope kept expanding without a forcing function. Lesson: if a project has no deadline and no first user, it doesn't ship. Either invent both or stop.