Datafact is an AI service for instant, audience-targeted survey data. You paste a Google Form URL, pick the respondent persona you want, and Datafact delivers realistic responses straight into your form. The focus here is applied AI engineering — turning a chosen persona and a form into coherent, human-like answers at scale. It runs on a fully serverless AWS architecture (Lambda, Step Functions, API Gateway HTTP APIs, and EventBridge Scheduler) for elastic, event-driven, no-ops execution. Live at datafact.site.

Avagenc is a multi-agent AI assistant that lets you run your day from a single chat interface. Through natural-language commands it connects to your real tools — Gmail and Google Calendar (read, compose, and manage email and events), Google Contacts, Tuya smart-home devices (lights, AC, outlets), and Spotify — and executes actions on your behalf, touching each account only when a request needs it. As co-founder I lead the agent orchestration layer: routing user intent to the right specialized agent, wrapping third-party APIs into reliable tool-calls, and shipping the product across web and Android. Privacy-first by design — no ads, no training on user data, no data selling. Currently in active development.

NusaVerify is an AI fact-checking engine that scores how likely a claim is to be a hoax. You feed it information — plain text, a URL, and more — and its AI searches across sources, then a validation algorithm with weighted averaging computes a validity / hoax probability. Every step of the AI's reasoning is visualized live as an animated mind-map, so the verdict is transparent rather than a black box. Built for the Bank Indonesia hackathon.