AI-Native Studio · Indonesia

Hello from nold.dev.We build AI-native appspeople open every day.

nold.dev is an AI-native product studio. Working with AI agents, we design, build and run apps for Indonesia and beyond: sports predictions, air quality alerts, tools for lecturers, and family health tracking.

See our productsGet in touch →
3 + 1
apps live, one coming soon
108
cities tracked by Kabut
4
sports in JadwalTanding
3
open-source Claude Code plugins
SCROLL
Portfolio

Our products

Each app solves one everyday problem well, and each one is built in-house, end to end.

JT

JadwalTanding

SportsLive

Predict the score before kickoff. Climb the leaderboard.

A score-prediction game for football, NBA, badminton and esports. Predictions lock five minutes before the match starts, then points arrive after the final whistle.

  • ›Exact-score and winner points, with a live leaderboard
  • ›Match intel: recent form, head-to-head and team averages
  • ›Crowd prediction split and a sports news feed
ExpoReact NativeTypeScriptSupabase
Visit jadwaltanding.com →
TodayLiga 1
Persib vs Persija
Locked · 19:00 WIB
2 – 1
Lakers vs Celtics
Open · closes in 42m
— : —
RRQ vs EVOS
Open · MPL ID
— : —
K

Kabut

Air qualityLive

Know when the haze is coming, and where it comes from.

Real-time air quality joined with satellite fire hotspots, so smoke can be traced back to nearby fires. Covers 108 cities in every Indonesian province, ranked worst air first.

  • ›Hourly sync from WAQI, Open-Meteo and NASA FIRMS
  • ›No login and no location prompt to start
  • ›Built to expand worldwide: cities are data, not code
ExpoSupabase Edge Functionspg_cronMapLibre
Visit kabut.net →
Worst air nowUS AQI
Pekanbaru
Riau · 14 hotspots within 50 km
182
Palangka Raya
Kalimantan Tengah
131
Jakarta
DKI Jakarta
96
Denpasar
Bali
38
ZD

Zonadosen

Lecturer toolsLive

Plan a curriculum and a semester workload without the spreadsheet chaos.

Two tools for Indonesian lecturers in one app. Papan Kurikulum lays out courses by semester and checks the rules before the accreditation meeting does. Perencana BKD estimates a lecturer workload plan and tells you if it passes.

  • ›Checks prerequisites, SKS limits per semester and minimum total SKS (Permendiktisaintek 39/2025)
  • ›Workload estimates follow the PO BKD 2021 rubric, with an Excel form ready to file
  • ›Imports Excel or CSV, exports PDF and Word. Data stays in the browser
Next.jsTypeScriptVitestClient-side PDF/DOCX
Visit zonadosen.com →
Papan Kurikulum145 SKS
Semester I
7 courses
20 SKS
Semester V
Over the 24 SKS limit
26 SKS
Data Mining
Prerequisite Basis Data is in a later semester
Check
L

Lullwren

Family healthComing soon

One calm app from cycle to pregnancy, birth, baby and recovery.

An all-in-one tracker for moms, dads, partners and caregivers. Private by default: data lives on the device.

  • ›Pregnancy week by week, contraction timer, kick counter
  • ›Feeding, sleep and diaper logs, WHO growth percentiles, vaccines
  • ›Cycle predictions and four postpartum recovery programs
ExpoTypeScriptSQLite + DrizzleZustand
Visit lullwren.com →
Pregnancy · Week 24Due 14 Feb
Baby size
Cantaloupe
30 cm
Kicks today
10 / 10
Next appointment
Thu 09:30
Client Work

We build for clients too

Web platforms, internal systems and site rebuilds, made with the same tools and workflow as our own apps.

Link-in-bio platform

Tautan

One link for everything a seller shares. Several pages per account, custom domains with automatic TLS, a live phone preview while editing, and click analytics that never store IP addresses.

Next.js 16PostgreSQLDrizzlednd-kit
Housing management system

Victory Park

Internal app for a residential complex: dues and finances, resident records, announcements, and CCTV access links, with separate admin and resident views.

Next.js 16SupabaseRow-level security
Redesign and performance

University website rebuild

A private university in Indonesia: we restored its WordPress site to a local staging copy, measured and cut slow queries, built a new standalone theme, and packaged a safe cPanel deploy.

WordPressDockerPythonPlaywright
How We Build

AI agents do the heavy lifting

Claude Code is part of how every product gets made. Agents write and review code, run tests, and research data sources, while we decide what to build and check every change before it ships. That is how a lean studio runs several products at once.

Plan

Spec with AI

Every feature starts as a written spec and plan, drafted with Claude and reviewed by us before any code exists.

Build

Agents write the code

Claude Code works across the codebase: Expo screens, Supabase migrations, edge functions and tests, in parallel when the work allows.

Check

Agents check, we approve

Agent code review, type checks and test suites run on every change. We approve what ships.

Share

We open-source our tooling

The Claude Code skills we write for our own work become public plugins for researchers and teachers.

One codebase per product

Every app ships to iOS, Android and the web from a single Expo and TypeScript codebase.

Lean backend

Supabase, Postgres and edge functions. No API keys in the client, cached reads, row-level security.

Built for local users

Bahasa Indonesia first, local leagues and cities, and apps that work on slow networks and respect privacy.

Open Source & Research

Open-source tools

Claude Code plugins we use in our own work, free for anyone to install.

Claude Code pluginacademic-writing-skills ↗Research integrity checks, bilingual EN/ID manuscript proofreading, and a journal-article workflow.Claude Code pluginresearch-methods-skills ↗NeuroKit2 biosignal and aeon time-series skills for scientific analysis.Claude Code pluginteaching-materials-skills ↗Course modules, textbooks and lecture slides aligned to Indonesian RPS, in English and Bahasa Indonesia.

Research

Academic work on cardiac signal analysis and fire forecasting, published as open code alongside the papers.

Research · Pythonpeatfire-eval ↗by Agustino HalimPeatFireBench: a district-month fire forecasting benchmark for Indonesia. 498 districts, 2012 to 2025.
Research · Pythonarrest-ecg-timing ↗by Agustino HalimTiming bias in ECG prediction of cardiac arrest: a matched case-control replication on the Nightingale arrest-ntuh-ecg dataset.
Research · Pythonaf-onset-within-patient ↗by Agustino HalimWithin-patient test of atrial fibrillation onset predictors, including a CPU port of the published AF-Mamba weights.
Research · Pythonsddb-warning-time-tests ↗by Agustino HalimPower and within-patient tests of HRV-based sudden cardiac death prediction on the MIT-BIH SDDB benchmark.
Research · Notebookscd-ecg-benchmark-audit ↗by Agustino HalimECG-based sudden cardiac death classification, re-audited with patient-level evaluation.
About

Meet the founder

Agustino Halim

Founder, nold.dev

Role
Founder and engineer
Builds
Every nold.dev product, design to deploy
Also
University lecturer in information systems
Research
Cardiac signal analysis, fire forecasting
GitHub ↗hello@nold.dev

Agustino started nold.dev to ship useful apps for Indonesia quickly, without a large team, by building with AI agents from the first spec to the final deploy.

Alongside the studio, Agustino lectures in information systems and does academic research on cardiac signal analysis and fire forecasting in Indonesia. The research is personal work, published as open code alongside the papers and listed under Research above.

Let's build something.

Projects, partnerships, or questions about our apps. We reply within two working days.

hello@nold.dev
GHJTK