Zeist

Creating things that should exist.

AmodalYour coding agents, working together.

PrimitivesInstruments for the fundamentals of design.

TimebaseVisualize and plan across time zones.

Outage ObserverLive status for providers in your stack. AboutWhy any of this exists. ArchiveEarlier work, eight of them, kept with what each taught us.

Zeist

Creating things that should exist.

AmodalYour coding agents, working together.

PrimitivesInstruments for the fundamentals of design.

TimebaseVisualize and plan across time zones.

Outage ObserverLive status for providers in your stack.

Swipe through. Tap to open.

The Arsenal, Venice, Giacomo Franco, 1610

Tool

Amodal

Your coding agents, working together.

Open amodal.co

The problem

Building with coding agents makes you the glue: copying handovers between Claude Code and Codex, reconstructing plans, picking a machine, reconnecting the same tools for each agent.

The approach

One workspace per project that holds the plan, the decisions and the memory. You direct the work in one conversation, on the web or in Telegram; it goes to the right agent on the right machine, and the results and the decisions that need you come back.

What we learned

The hard part is not running agents but continuity. The plan, the decisions and the exact code checkpoint have to travel with the work, or every switch starts from zero.

Paired machines with authenticated runners, then safe dispatch and results, Claude working alongside Codex, and Telegram. Then an early-access alpha.

Status: Nearly there; the waitlist is open
Runs on: The web and Telegram, with a runner on your own machines
Since: September 2026
Built with: A hub on Cloudflare; runners on your own Mac or Raspberry Pi drive Claude Code and Codex

The Arsenal, Venice The gate of the city's shipyard, where thousands of specialists built ships together, stage by stage. Giacomo Franco, 1610.

Tool

Primitives

Instruments for the fundamentals of design.

Open primitiv.es

The problem

Everyday design questions, like whether two colours are readable together or what a type scale steps to, are answered by scattered one-off sites. Each has its own layout, its own rounding, and rarely an answer you can send to someone.

The approach

One instrument per question, grouped by primitive, in one harness: the same controls, the same explanation and a stateful address everywhere. The maths sits in one package, JSON in and JSON out, so agents can later ask the same questions people do.

What we learned

The harness is most of the work; once it was right, a new instrument became a folder. Precision is mostly rounding: a colour written to the address has to come back as the same colour, or a contrast check passes on one screen and fails on the next.

Grid, shape and motion in turn. A public requests board orders what comes after, and the maths opens up as endpoints for agents.

Primitives on a phone — primitiv.es, as it is today.

Status: Live; Color and Type are out, eleven instruments
Runs on: The web
Since: September 2026
Built with: Astro and Svelte on one Cloudflare Worker, D1 for requests; the maths in one package, JSON in and JSON out
Also: GitHub

Paestum, Italy Greek temples about 2,500 years old, down to the column and the beam. Piranesi, 1778.

Royal Observatory, Greenwich, John Varley, about 1830

Tool

Timebase

Visualize and plan across time zones.

Open timebase.me

The problem

Every world clock shows the same row of cities and the same offsets. None of them tells you when the people you work with will be awake, or when the thing on your calendar happens in their day.

The approach

A row per city and colour for the hour: a continuous 24-hour gradient carries the time of day, so icons, terminators and offsets could go. One scrub gesture moves every clock at once. Countdowns to the events on your calendar, tinted by the hour where they happen.

What we learned

Colour can carry what labels usually do; once the rows carried the hour, everything else could be removed. Building the iPhone app and the web app as separate, bespoke things was faster than sharing code, and both are better for it.

iPad, Mac and watch, each designed on its own, after Shio ships.

Timebase on a phone — timebase.me, as it is today.

Status: Live on the App Store and at timebase.me
Runs on: iPhone, and the web
Since: 2026
Built with: Swift and SwiftUI on iOS; plain HTML and JavaScript on the web, no backend
Also: App Store, GitHub

Royal Observatory, Greenwich The prime meridian runs through its courtyard. Every time zone is counted from this line. John Varley, about 1830.

Eddystone Lighthouse, After Henry Roberts, 1761

Tool

Outage Observer

Live status for providers in your stack.

Open outage.observer

The problem

Every provider has a status page; nobody watches 151 of them. An outage reaches you as your own error rate, or as a post on X, before the page admits it.

The approach

One Worker polls the official status pages of 151 infrastructure and AI providers, detects each transition exactly once, and tells you where you already are: a bot in Telegram, Slack or Discord, a push, an RSS feed, the public board, or a Mac menu-bar app.

What we learned

The hard part is not polling; it is not lying. Atomic transition detection, alerts that know which region you are in, and self-hosted brand icons so nothing phones home. Bot Fight Mode blocks your own heartbeat if it hits the custom domain.

Per-provider pages with uptime history.

Outage Observer on a phone — outage.observer, as it is today.

Status: Live; launched on Product Hunt in June 2026
Runs on: The web board; Telegram, Slack and Discord bots; push, RSS; a Mac menu-bar app
Since: June 2026
Built with: One Cloudflare Worker with D1 and KV, on the free tier; a native Swift Mac app
Also: Telegram, GitHub

Eddystone Lighthouse Built on a reef off Plymouth in 1698, to warn ships about what they could not see coming. After Henry Roberts, 1761.

Zeist

Zeist

Amodal

The problem

The approach

What we learned

Next

Primitives

The problem

The approach

What we learned

Next

Timebase

The problem

The approach

What we learned

Next

Outage Observer

The problem

The approach

What we learned

Next