Paperbird

— a creative AI studio

A case study · Private beta

Hush

A quiet workspace for noisy teams.

Year

2025

Status

Private beta

Role

Product direction, design, engineering

Client

Independent product, Paperbird

AI Agents

Product Design

Next.js

Editorial UI

Open notebook on a concrete surface catching morning light

Fig. 01 — Hush, studio shot — 2025

Chapter I

An overview

Hush started from a simple frustration: the teams we admired were drowning in notifications while starving for attention. Slack, email, Linear, Notion — each a river, each demanding presence. Hush is our answer. It reads across those streams and composes a single, daily letter: what changed, what's waiting, what to read.

Margin

Every case study on this site is written in full sentences on purpose. If we cannot explain the project as prose, we do not understand it yet.

The best software today feels like a letter. It arrives, says one thing clearly, and closes itself.

Rafael Ortiz, CPO

Chapter II

The challenge


Modern workspaces treat every message as urgent and every feed as infinite. The cost is quiet work — the kind where hard problems actually get solved. We needed an agent that could hold the chaos without inheriting its shape.

Chapter III

Our approach


Rather than building another dashboard, we designed Hush as an editor. Every morning, an agent reviews the last 24 hours across connected tools, discards the forgettable, and drafts a brief. Users read it the way one reads a newspaper — start to finish, once.

A pair of hands writing in a notebook beside a laptop

Daily brief prototype — early reading tests

Dark desk with a single illuminated screen at night

Agent console — internal view

Close-up of typography on a monitor

Typography study — reading column

Long horizontal highway cutting through wheat fields at dusk

Field research photograph — quiet as a feature

Chapter IV

The solution

We paired a retrieval layer across a team's stack with a writer model tuned for restraint. The interface is a single reading column — serif, 68 characters wide, no sidebars — so the act of reading Hush feels like the opposite of checking a feed. When the user is done, they are done.

Chapter V — Outcomes

At a glance

01Time in tools reduced

41%

02Clarity on priorities

34→78%

03Daily brief open rate

94%

04Teams in closed cohort

22

In a closed cohort of 22 product teams, Hush reduced time-in-tool by a median of 41% within six weeks while lifting self-reported clarity on priorities from 34% to 78%. Private beta opens Q3 2026.

Next in this issue

Dossier

A notation for building with AI, spoken plainly.