Problems/Lab
In the labWorkbench

Real problems. Working solutions.

Not finished case studies. Working answers being built and tested.

Each one exists because the available answer was too complicated, too passive or aimed at the wrong result. Status is stated as it is: building, testing, or waiting on the next proof. Nothing here claims a validated result yet.

P-01My Health Pulse / LookiBuilding

The data existed. The useful signal did not.

ProblemPassive observation can create a huge personal record without showing what changed. Most dashboards then mistake missing coverage for inactivity, bury the person in charts or drift into diagnosis.
Working solutionA calm monitor that archives Looki observations, states coverage honestly and compares behaviour with the person's own baseline.
Current buildResponsive prototype running; cumulative data architecture still being hardened.

Next proof Can it detect a meaningful deviation without false reassurance or false alarm?

Working model

ObservationsCoverageOwn baselineChangeHonest unknown

* Boundary: never diagnosis.

P-02Exile SageTesting

The game gave the player more information than a player could use.

ProblemA player has to judge items, build trade-offs, market value and progression while playing. Guides arrive before or after the moment that matters: what should I do next?
Working solutionAn iPad/second-screen companion that reduces the game state to one next-best action, item/build fit and remembered decisions.
Current buildResponsive and ultra-wide prototypes running; live observation not solved yet.

Next proof Can it improve the next decision without interrupting play?

Working model

Game stateRelevant signalsOne next action

* Boundary: recommend, never automate.

P-03WypsterTesting

The website looked wrong. The business proposition was unclear.

ProblemBusinesses request a new website or SEO fix when the proposition itself is unclear. Conventional audits begin with pages and keywords, then produce a long report nobody acts on.
Working solutionStart with positioning, offer, value and pricing; ask a person to approve the diagnosis; only then assess content/UX/SEO and generate a new direction.
Current buildPrototype plus its research component exist; diagnosis quality being compared.

Next proof Is the diagnosis consistently better than a generic automated audit?

Working model

DomainPropositionHuman agreementSite direction

Also on the bench

  • P-04AddyssBuilding

    The ad platform optimised the campaign. The business target came second.

    Paid-media tools open with budgets and keywords, so the business number gets fitted around the media plan. Addyss starts at the target and calculates feasibility and budget backwards.

    Current build — Backwards planning model in progress.

    Next proof Can backwards planning produce a budget useful enough to guide a real campaign?

    Working model

    Business targetFeasibility + budgetDeviation → decision
  • P-05Hoshin BuilderNext proof

    The strategy existed. The links between priorities, owners and actions did not.

    Strategy matrices produce a document while the relationships stay in someone's head. The builder guides the links between aim, priorities, actions, measures and owners.

    Current build — Guided linking flow drafted; follow-through not yet observed.

    Next proof Does it improve follow-through, or only produce a cleaner matrix?

    Working model

    AimPrioritiesActionsMeasures / owners
The thinking behind theseTalk about one of these