Menu
HomeSignal Flow

Current R&D · Architecture exploration

Signal Flow

Signal Flow explores how incoming information can be understood step by step and turned into structured cases, decisions and actions — using AI where interpretation is needed and deterministic software where rules can take over.

Conceptual architecture

Learning reaches back into the work.

Follow a case from an incoming signal through review and permitted work to learning feedback.

Signal Flow — layers and learning feedbackA Sentinel role inspects the arrival at the Sense boundary. The cyan case moves through Interpret, human review at Decide and a permitted internal record update at Act, then enters Learn. Green feedback paths reach Act, Decide, Interpret, Sense, Purpose and Govern. They show where learning can inform review, not a claim that every case automatically changes every layer. Purpose and Govern remain context. This is a synthetic architectural illustration.PURPOSECompany contextScope · vocabulary · goalsSENSEFind the concernINTERPRETMake meaning explicitDECIDEPropose the recordACTCarry out permitted workLEARNImprove from experienceGuidance · specificationsGoverned settingsGOVERNSentinelARRIVALCASE ARoutine recordExperienceScreeningInterpretingPerson reviewingRecord updatedExperienceLearning feedbackPerson-reviewed recordLearning feedbackFor review across the architecture
Signal Flow — layers and learning feedbackA Sentinel role inspects the arrival at the Sense boundary. The cyan case moves through Interpret, human review at Decide and a permitted internal record update at Act, then enters Learn. Green feedback paths reach Act, Decide, Interpret, Sense, Purpose and Govern. They show where learning can inform review, not a claim that every case automatically changes every layer. Purpose and Govern remain context. This is a synthetic architectural illustration.PURPOSECompanycontextSENSEINTERPRETDECIDEACTLEARNGuidanceSpecificationsGoverned settingsGOVERNSentinelARRIVALCASE ARoutinerecordExperienceScreeningInterpretingPerson reviewingRecord updatedExperienceLearning feedbackReviewedLearning feedbackFor review across the architecture
A conceptual lifecycle with Learn feeding back to every preceding layer. The flow adapts to the work; not every case follows a single, rigid pipeline. Learning can inform every layer. Changes follow their own review and adoption rules.

Signal Flow board

Work keeps moving.

New cases enter. Completed work leaves. Other work waits or runs between steps.

Ten synthetic cases occupy nine vertical columns, each four cards high, and their in-board gutters. New arrivals replace completed work one at a time. Several clear routes move together. A few cases pause for person review of Lane, show acceptance, then continue into Intent. Every card eventually leaves; later routine record work follows its permitted path. Counts illustrate completed calls in selected routing turns, not live telemetry.SENSEINTERPRETDECIDEACTLEARNSplitRouteLaneIntentDatanormalizationMatterDecideActDreaming• Work under wayPerson reviewResults in columns · work between steps
Three synthetic cases enter and leave a focused Lane-to-Intent view. Each column is four cards high; waiting and running work stays between the columns within those rows. A few cases pause for person review. The person accepts Lane, the card resumes into Intent, and every card eventually leaves the view.INTERPRETLaneIntent• Work under wayPerson reviewLane → Intent · selected steps

AI Architecture Notes

Read the Signal Flow series

  1. 02 / AI Architecture NotesGive the work an identityRead the article
  2. 03 / AI Architecture NotesGive the decision somewhere to goRead the article
  3. 04 / AI Architecture NotesWhat should change after a correction?Read the article
  4. 05 / AI Architecture NotesStart with the work you already haveRead the article

Time Machine

Start with the work you already have

Time Machine brings past work into view. People and AI use those examples to shape the vocabulary, useful records and application connections.

A synthetic onboarding method around Signal Flow’s Time Machine. Historical arrivals remain available while a person and AI explore and review a branching workflow, lane and intent vocabulary. A trial case gains normalized facts; two related cases connect to one continuing matter. A meaningful bucket proposes a record that is matched to an existing application contract. A separate unfinished bridge and application show design work still needed. Application contracts can inform the design. The map remains visible as the foreground sequence repeats.Earlier workPastHistory retainedOrder ADelivery movedto ThursdayPrepare togetherAIPersonMeaningEvidenceRecordsCapabilitiesDraft and reviewReviewedA map of the workWorkflowLaneIntentTry one caseWith the reviewed mapNormalized factsCASE AOrder ADate · ThuRelated casesCASE BOrder AOne matterOrder ADelivery changeOrder updateExisting APIBusinessapplicationBuild what is missingLet capabilities shape the designOrder A
A synthetic onboarding method around Signal Flow’s Time Machine. Historical arrivals remain available while a person and AI explore and review a branching workflow, lane and intent vocabulary. A trial case gains normalized facts; two related cases connect to one continuing matter. A meaningful bucket proposes a record that is matched to an existing application contract. A separate unfinished bridge and application show design work still needed. Application contracts can inform the design. The map remains visible as the foreground sequence repeats.Earlier workPastHistory stays availableExplore and reviewDraftmapReviewitReviewed definitionsChoose a branchWorkflowLaneIntentTry one caseNormalized factsCASE AOrder ADate · ThuCASE BOrder AMatterDeliverychangeExisting APIApplicationBuild whatis missingRefine the designOrder A
Review the definitions, then try them on a case. Shared identities connect related cases to a matter. Useful records meet existing application contracts—or reveal what needs to be built.Read Article 5

Does your system need a next step?

I’m available for selected consulting engagements, primarily part-time.

Discuss an engagement