Skip to content

Grey areas, made sharp

Sharp grey is a software studio. Instead of renting a SaaS that only half fits, get software built around how you work, on modern web technology. You own what we build. Run it yourself, or let us run it for you.

Start a project See the work

Selected work

All work
IWE’s editor: a chapter with its findings marked in the text, beside a prose quality score of 82 and its breakdown.

Our product · Live

IWE

An editor for long manuscripts. 38 analyzers read your prose as you type, and 3 AI tools coach, draft and edit beside you.

Fetchpawz’s new identity on a business card, a push notification, a pet’s swipe card, the Top Cat rosette, the app icon, collar tags and stickers.

Partnership · Invite-only beta

Fetchpawz

The pet app with a rescue mission. Owners rate each other’s pets, and rescues list adoptable animals in the same deck. We build it, designed its new identity, and co-own it with its founder.

The 3D configurator: a navy rolling shutter in 3D, beside the opening, mounting and curtain series options.

Client · In development

3D configurator

For Pentagon Security Shutters. A customer builds a rolling shutter in 3D, sees it on a photo of their own building, and gets a spec sheet for the dealer.

Pocket Philosophy’s link preview: the logo, the line “One question a day. Two AIs argue it out. You find your place in the grey.” and a crowd of dots over a scale.

Our product · In development

Pocket Philosophy

One question a day. Two AI models from different labs argue it out, and you map where you stand.

The work in numbers

  • 164

    faults planted in three stories written badly on purpose, which IWE is tested against so it can’t quietly stop catching one

  • 500 m

    the radius inside which a missing pet’s public location is drawn at random, so no Fetchpawz page leads to a door

  • 32,256

    shutter configurations the 3D configurator’s audit rebuilds to prove that nothing clips

  • 0

    player answers that can reach a model: Pocket Philosophy’s engine can’t read the players’ tables

What we build

What we build, in full

  1. Tools your customers use themselves

    Configurators, calculators and visualizers on your own site. The customer sets up the job, and your team receives a complete specification. Example: the 3D configurator for Pentagon Security Shutters.

  2. AI features and products

    Models that draft, score, sort or answer, working on your own content. Tested before they ship, scored after. Examples: IWE and Pocket Philosophy.

  3. Web apps and platforms

    A product for your customers, a tool for your team, or a marketplace with two sides, with the accounts, payments and admin built properly the first time. Examples: IWE and Fetchpawz.

  4. Product branding

    A name, a look and a voice for your product, designed with it and built into its code, so the brand can change later without touching a feature. Examples: the new Fetchpawz identity, and Tenterhooks, a product for writers built on IWE.

  5. Launch and hosting

    Every launch comes with tests, monitoring and written procedures. Then you choose: host it yourself with everything handed over, or let us run it for you.

How we work

How we work, in full

  1. Numbers, not adjectives

    We tell you what a thing does and how well. If it’s still in development, we say so.

  2. Measured AI

    An AI feature has to pass a set of test cases before it ships, and we keep scoring it after.

  3. Built to run

    Whoever hosts it, it comes with tests, monitoring and a plan for the failures we can see coming.

  4. Show the workings

    You see the reasoning, the trade-offs and the test results behind your project, not just the finished thing.