Current experiments

What I’m messing about with.

I normally have a few things running alongside the paid work: products, frameworks, research questions and the occasional system that began with “surely this could be less annoying”. This is the rolling view of what I’m building, testing, deliberately pausing and trying to understand.

Some collaborations are deliberately described at a higher level. The useful story is the problem, the product thinking and what is being learned. Other people’s names and businesses are not raw material for my content.

On the workbench

Products, systems and questions in various states of becoming real.

Building now

The projects receiving the most active poking, building and argument.

Some are already live. Others are becoming a body of work sturdy enough for somebody else to use. All of them are trying to turn a complicated system into something more useful.

01
Product frameworkVolume IV

Active

Signature Experience Platform

An AI-native, mobile-first product framework and reference architecture for creating deeply personalised, trusted and continuously orchestrated relationships between an individual and an organisation.

It asks what personalisation becomes when it stops meaning campaigns, segments and recommendations, and starts behaving more like a relationship with memory, context, consent and intent.

Current goal

Complete the Mobile Experience Blueprint through human service, explanation, control and consequential action, then keep turning the wider six-volume foundation into something teams can actually design and govern against.

02
Live productNative SwiftUI parity

Active

DunDunSee™

A redundancy package helper built because the moment somebody needs to understand notice, tax and runway is generally not the moment they want to become a spreadsheet enthusiast.

The web product is live. The current experiment is whether its behaviour can be carried into a proper native iOS implementation without losing the small interaction and lifecycle decisions that make it useful.

Payments are not hooked up yet, so play as much as you like.

Open the live product

Current goal

Validate the current SwiftUI source-parity marker with compile, XCTest, simulator and physical-iPhone evidence, then close the remaining visual, accessibility and lifecycle gaps before deciding whether native parity has actually been earned.

04
Discovery methodRecovery & canon

Active

Platform Archaeology

A method for recovering the real shape of a digital platform from the debris organisations leave across tickets, documents, code, spreadsheets, old roadmaps and the memories of people who happen to still be around.

The point is not to summarise a folder. It is to reconstruct the systems, decisions, ownership, duplication, contradictions, risks and missing knowledge around the work.

Open the Coin Cascade prototype

Current goal

Turn the source-recovery and canon-building techniques into a repeatable professional method for understanding an estate properly before somebody attempts to automate the lot.

12
Living story worldBook One canon

Active

Clenchmere

A long-form literary world that began in conversation and is now being run as a version-controlled creative system: source material, accepted canon, characters, places, continuity, experiments and manuscript work all separated properly enough that the story can grow without forgetting itself.

The recovery phase is now doing its job rather than being the whole project. Book One Movement II is canonical through Chapter 16, while the same source-of-truth model continues to support maps, dossiers, reader experiments, audio, imagery and other ways into the world.

Visit the Clenchmere site

Current goal

Carry Book One through the Chapter 18 readiness boundary, test whether the recovered canon is strong enough to support formal manuscript production, and keep presentation growing from truth rather than racing ahead of it.

13
Engineering platformPrivate prototype

Active

Quality Platform

A private experiment in whether a distributed engineering organisation can turn local problem-solving into shared capability without centralising innovation or making a central Quality function own everybody else’s testing.

The working hypothesis is simple: do not centralise innovation; centralise the mechanism that allows successful innovation to spread. The interesting work is therefore as much about sponsorship, evidence, promotion and ownership as it is about tests.

Current goal

Prove named identity and persistent-write security end to end, then implement the first protected flow from problem and experiment proposal through evidence and promotion into a reusable capability.

14
Incremental gamePlayable v0.3

Active

Cranes That Crane Cranes

An incremental browser game with an unnecessarily self-explanatory premise: stronger cranes crane the cranes that came before them.

What began as a tiny playable loop has grown through greyscale toy machinery, industrial cranes, orbital infrastructure and a reality-bending gantry into a full Collapse cinematic that rewinds the whole thing back to the bedroom where it started.

Current goal

Turn Collapse from an ending into memory: capture what each run became, preserve it as a permanent universe, then start exploring what happens when many completed universes keep contributing to a larger multiverse.

15
Worldbuilding systemFoundation pass

Active

GASA

The Galactic Astronomical Survey Authority is a worldbuilding experiment about what happens to human mythology, identity, navigation, law and culture when we travel far enough through space for the familiar constellations to distort and disappear.

The setting is being built as a universe first and a game second. Travel itself can change what civilisation knows: some routes may mirror a universe, others may bend time toward the birth of constellations, and apparent steps forward or backward can both become progress because they reveal something new.

Current goal

Build the foundations of Travel, Space and Time before getting seduced by game mechanics: define the strange routes, risks, consequences and discoveries that make moving through this universe meaningful in the first place.

Taking shape with others

Private collaborations, described without borrowing somebody else’s identity.

These are real product and platform problems being worked through with other people. The public summaries stay deliberately focused on the system, the goal and the learning.

06
Partner platformEarly product work

Pending client review

TAI

A low-overhead data, orchestration and transaction platform exploring how fragmented information, manual decisions and repetitive hand-offs can become one controlled workflow.

The interesting bit is not simply gathering more data. It is connecting evidence, judgement and action without building an organisation whose main product is moving information between spreadsheets.

Current goal

Define the data model, automate the routine movement from evidence to action and keep meaningful approval with a human.

07
Partner productArchitecture & discovery

Pending client review

A joined-up coaching platform

A secure product for a small coaching business, bringing intake, client records, programmes, progress, communication and payments into one manageable service.

The work is less about recreating every existing tool and more about deciding which parts should genuinely belong together, where privacy matters and how the service can grow without becoming operational soup.

Current goal

Set the architecture, roles, privacy model, cost boundaries and delivery sequence before replacing a collection of disconnected tools.

16
Partner website experimentPositioning & proposition

Pending client review

A specialist B2B website

A specialist B2B website experiment focused on making deep subject expertise understandable before asking a visitor to learn the machinery and jargon underneath it.

The work reframes the proposition around the problem, the value and a clearer route through the detail, while keeping the specialist depth available for people who actually need it.

Current goal

Complete collaborator review of the proposition and working site, then settle the production configuration and handover without exposing private commercial or client detail.

17
Partner experience experimentInteractive studio site

Active

A creative studio experience

A creative-studio website experiment where the page itself acts as part of the demonstration rather than simply describing the work.

Scrolling moves through layered landscapes toward a distant summit, ending with a small pixel character carrying and planting the studio flag. The useful question is whether motion, depth and pacing can communicate the proposition without turning the site into a technical circus.

Current goal

Keep refining the interaction as a lightweight, responsive demonstration while preserving the collaborator’s privacy and keeping the experience sturdy across real devices.

Elsewhere on the workbench

Not everything needs to become a business model.

Some things deserve a proper home simply because losing useful work to old platforms and forgotten folders is a particularly dull kind of waste. A paused project stays here when the work is real and the next move is still worth remembering.

03
Public researchLocalisation pilot

Paused

The Six-Year-Old Manifesto and Test

A public test for whether advanced intelligence expands ordinary human capability, security and freedom.

The first published run now also has an isolated Spanish localisation pilot, testing whether the experience can travel between languages without duplicating or quietly changing the canonical evidence underneath it.

Open the live manifesto

Current goal

When the experiment resumes, review the Spanish experience technically and editorially, test it on physical devices with native-language input, then decide the long-term localisation architecture before using Arabic and Japanese as harder stress tests.

05
Original gameDesign canon

Paused

The Department of Impossible Logistics

A procedural management game about running a shabby civil-service department responsible for problems that should not technically exist.

Cases arrive. Policies accumulate. Directors approve, departments request and workers execute. Small decisions become institutional behaviour, usually with consequences nobody quite ordered.

Current goal

When development resumes, turn the substantial design canon into a playable vertical slice that proves the central loop of noticing something, making a decision and watching the institution respond.

08
Creative recoveryArchive concept

Paused

DiscoMcDisco Music Archive

A lightweight home for music scattered across an old community site, personal recordings and newer generative experiments.

The ambition is intentionally modest: a clean playlist, simple player controls and a visual backdrop. Not Spotify in a shed.

Current goal

Recover and organise the catalogue, preserve what still works and rebuild it as a small standalone archive on the DiscoMcDisco domain.

09
Live interface experimentSession only

Active

Styling Easter Egg

Four portraits escaped containment and became complete visual systems. These are not polite colour themes. They change typography, cards, borders, spacing, buttons, imagery and the general temperament of the entire site.

Choose one and the change happens immediately. It follows you around this site for the rest of the browser session, then politely forgets everything when the session ends. Original always puts the furniture back.

Original site styling is active. Experimental styles are remembered only for this browser session.

Current goal

See how far a content-led site can change its visual identity without damaging meaning, navigation or usability, while keeping the whole diversion reversible and session-only.

10
Household productRecipe verification

Active

Dusty’s Favourite Foods

A lightweight, searchable household recipe collection for Russ and Soph, with ingredient search, filters, sorting, recipe views and source indexing without reproducing cookbook content.

The current work is less about adding more recipes and more about whether the collection can prove what it knows: structure, servings, calories, macros and reusable components checked against a strict, versioned Gold Standard rather than a vague confidence score.

Open the live product

Current goal

Settle the Gold Standard rules, run a representative read-only pilot across 10–20 recipes, then decide what deserves a full-dataset pass before any automated correction is allowed near the food.

11
Playful productAcceptance candidate

Paused

Cheesy Playlist

A phone-first music experiment where somebody describes the night, the product generates a playlist recipe and the record-shop scene changes to match the atmosphere.

The fixture-driven journey and generated-wallpaper path are built. The remaining question is wonderfully less theoretical: does the real image-generation flow behave properly when it is actually funded and used on a phone?

Open the live product

Current goal

When the experiment resumes, configure the preview image key, run one real generated-wallpaper journey and complete physical-iPhone acceptance before deciding whether the review branch is ready to become the product.

A useful pattern

Most of these began because a real system felt more complicated, wasteful or unfair than it needed to be.

That is usually a reasonable place to start a conversation. Particularly when your problem rhymes with one of these, even if it wears a different badge and lives in a different organisation.

Start a conversation