$ ./crafted-in-context --status build.0001 · 2025-q4

Agent-first
engineering,
shipping weekly.

// readme

We build software for the next decade of work — where the user is sometimes a person, and increasingly an agent acting on their behalf.

Products live
02
dayora · testimark
Loop time
< 1 wk
idea → users
./view products → ./contact --reason=partnership
§01 / what we do three.modes
mode_01

Vertical AI products

We pick a real workflow and own the entire context around it — from data, to model, to the surface a user (or their agent) actually touches.

mode_02

Agent-first engineering

Every system we ship is designed to be driven by both humans and agents. If an agent can’t use it, neither can scale.

mode_03

Automated product loops

Our agents read what users actually do, decide what to try next, and ship the iteration themselves. The team stays in the taste seat; the loop runs on its own.

§02 / principles.md

Five rules
we work by.

Versioned, opinionated, designed to be argued with. We rewrite them in public when they break.

// principle.01 01 / 05

Context is the product.

Models are commodities. The thing that turns a model into useful work is the context around it — data, tools, memory, taste. That is the artefact we craft.

// principle.02 02 / 05

Engineer with agents, not around them.

Agents are first-class teammates in our process. They write code, run experiments, read user behaviour and propose the next change. We design the rails; the agents do the laps.

// principle.03 03 / 05

Self-iterating products.

Our agents understand the product, the users and the gap between the two. They generate hypotheses, ship trials, and roll back what doesn’t work — without waiting for a human standup.

// principle.04 04 / 05

Vertical, not horizontal.

We don’t build platforms. We build deeply opinionated tools for specific workflows, then point our agents at the seams between them and let composability emerge.

// principle.05 05 / 05

Humans hold the taste.

Agents propose, ship and measure. Humans decide what is good. The studio’s job is to keep that line bright while the rest of the loop accelerates.

§03 / products

Two live.
More to come.

02 in market · roadmap = private
● Live · 2025 dayora.ai ↗

Dayora

A free AI journaling app with mood tracking, voice entries and daily insights — built around a quiet conversational agent that learns what actually moves your week.

Wellbeing Conversational Voice + text
● Live · 2025 testimark.co ↗

Testimark

Agentic evaluation for software teams shipping AI-driven products. Turn fuzzy quality into a measurable, regression-tested loop.

Eval Developer tools Agents
§04 / about.txt

Small studio.
Long loops.

› Crafted in Context is a small Australian studio. We design, engineer and operate a focused portfolio of agent-first products.

› We are not an agency. We don’t take outside briefs. Everything we ship is something we use, would pay for, and intend to run for a long time.

› The work is opinionated, the team is small, and the loop between writing code and watching a real user (or their agent) use it is measured in hours.

§05 / company.json verifiable on record 
{
  "legal_name": "Crafted in Context Pty Ltd",
  "acn": "694 376 853",
  "jurisdiction": "Australia",
  "founded": "2025",
  "contact_email": "jdelisle@rsq.net.au",
  "disciplines": "product, engineering, design, ops"
}
§06 / contact

Open
a thread.

Form below opens your mail client with the right subject and address. Plain text, no auto-replies — you’ll hear back from a human inside 48 hours.

direct line
opens your mail client