Skip to main content

Phygital Mycelium

The unseen network that makes everything happen.

Mycelium is the underground network. Mushroom caps are the ventures that pop up wherever conditions are right. PRDs describe capabilities — shared primitives that any venture can compose from. Today there are seven mushroom caps. If this works, there will be millions.

Business Factory — Stackmates

Capabilities

Each PRD describes a Mycelium capability — a shared lib that any mushroom cap can use. The capability owns the job definition and success criteria. The venture owns the business model and the customer.

Active

Priorities: Table order = build order.

Top row is what the engineering team builds next. Blocked platform items go first — they block everything downstream.

PRDCapabilityCurrently Growing InStatusNorth StarKill Date
Identity & AccessMulti-tenant auth, RBAC, org isolationPlatformBLOCKED — owner locked out (22P02)Owner logs in, no manual DB
Sales CRM & RFPDemand-driven CRM + RFP workflowstackmatesWire algorithms — blocked by IdentityWin Rate > 30%2026-03-24
Agent PlatformIdentity, memory, comms, dispatch, VVFLPlatformWire ETL pathAgent productive in under 30s
Sales Dev AgentAI-powered demand generationstackmatesBuild + wire5 discovery calls/month2026-03-24
Sales Process OptimisationProcess intelligence + agent team for salesstackmatesSpec1st customer via agent-optimized process
ETL Data ToolUniversal data readiness platformstackmatesShippingQuery Latency < 2s
AI Quality Gates90/10 AI/Human production protocolPlatformActive80% Gross Margin on services
Agency LayerAI-Native Agency (90/10 split)Shared MyceliumActive80% Gross Margin on services
Content AmplifierMulti-platform content distributiondreamineeringWire workchartsMulti-platform dispatch >= 60%
Data InterfaceThree-interface data accessstackmatesSpec draftEvery table accessible via Admin + Chat + A2A
Prompt DeckAlignment instrument (5x5)dreamineeringSpec + prototypeD7 return rate > 35%
Make Meetings MatterStructured meeting state changesstackmatesSpec draft>= 90% meetings produce state change
Sui Wallet Safety PatternsWallet safety standardsstackmatesSpec completeChain-agnostic safety standard adopted
Time + Mind NavigationIntention/Attention calendardreamineeringSpec readyDreamineering Yield > 0.5 art/hr
Trust CommerceSui-based coach-selling + on-chain reputationShared MyceliumSpecReferral conversion >15% (vs 2-3% cold ads)
DePIN InfrastructureMulti-chain device + data + rewardsstackmatesSpec draftCross-layer workflow executes end-to-end

Backlog

PRDCapabilityCurrently Growing InDifficulty
Commissioning State MachineProgressive data maturity automationstackmatesMedium
Drawing ToolA&ID visual language primitivesdreamineeringMedium
Workflow EngineDecision engine for work receiptsstackmatesAdvanced
Social Media MarketingLinkedIn-first B2B automationdreamineeringMedium
Workflow BuilderVisual workflow editorstackmatesMedium
Sleep protocol content pagePractice designbetterpracticeStarter
Morning intention practicePractice designbetterpracticeStarter
Optimism daily protocol trackerFeedback loopsbetterpracticeStarter
Builder onboarding flowUX, empathystackmatesMedium
Trail recommendation engineData, AI, productberleytrailsAdvanced
Mental model searchKnowledge architecturedreamineeringAdvanced

AI-Native Production

We operate with software economics in a service-delivery world.

PrincipleProtocolGoal
90% AI ProductionAgents draft, research, and format all initial deliverables< $60 cost per unit
10% Human QA20 mins of "taste and judgment" review per unit0% defect rate
Outcome PricingPrice by deliverable, not by hour> 70% Gross Margin

This protocol is documented in the AI-Native Agency playbook.


Commissioning Dashboard

One view of every capability — what's specified, what's built, what's proven. Maturity levels from L0 (spec only) to L4 (independently verified). Flight readiness gates. Kill criteria. Revenue readiness.


Product Layers

Each PRD must define the minimum feature set needed to run one complete fair-trade loop: question -> problem -> trade -> proof -> feedback.

Product LayerMust-Have FeaturesWhy It ExistsPrimary References
Horizontal SaaSAuth, CRM, pipeline, tasks, documents, analytics, workflow automationShared operating surfaces across all venturesSaaS Toolkit
Vertical SaaSIndustry workflows, role-aware entities, compliance and domain scoringConvert generic tools into outcome systems by industryVertical RaaS + Industries
TokenizationOwnership model, transfer model, settlement modelMake value and contribution portable and verifiableTokenization
TokenomicsIncentives, distribution, governance, retention loopsKeep participation fair, sustainable, and anti-extractiveTokenomics
Phygital ExtensionDevice identity, agent identity, task attestationSupport humans, AI agents, and devices in one loopDePIN

Feature Gate

Before a PRD is considered build-ready:

  1. Workflow proof: One end-to-end user workflow with measurable before/after.
  2. Entity proof: Core entities and links are explicit (who, what, where, value).
  3. Loop proof: Feedback metric exists and updates behavior.
  4. Business proof: Pricing and acquisition assumptions are testable this cycle.
  5. Trust proof: Safety and settlement model are explicit (especially if tokenized).

The Factory

Stackmates for Dream Engineering

PRDs are the specification layer. The engineering repo is the execution layer. The team that builds it never validates it.

DREAM REPO (this)                    ENGINEERING REPO (stackmates)
───────────────── ─────────────────────────────
1. Write PRD spec → RFP → 2. Architecture + build
3. Ship feature
4. Commission (validate) ← DELIVERY ← "Ready for inspection"
Agent reads spec
Agent opens browser
Agent checks features,
functions, outcomes
against PRD criteria
5. Pass / Fail + evidence
StageWhereWhat Happens
SpecHere — /phygital-mycelium/Define features, functions, outcomes, success criteria
MethodologyHere — /docs/products/JTBD, Shape Up, Flow Engineering, Type-First
ArchitectureEngineering repoDB plans, schemas, service boundaries, implementation
BuildEngineering repoCode, tests, deployment
CommissionHere — PRD commissioning tableIndependent verification against spec

In factory engineering, the team that builds a system is never the team that commissions it. The builder knows what they intended. The commissioner checks what actually shipped. A seed phrase got deleted because nobody independently verified. The Sui Wallet PRD exists so that won't happen again.


Build-to-Revenue Sequence

PRDs are not product wishlists. They are business development instruments.

PhaseProduct MoveBusiness Dev MoveOutput
1. SpecDefine jobs, features, success criteriaDefine ICP and offerPRD complete
2. PrototypeShip smallest usable loopRun first 5 discovery callsFriction map
3. PilotSolve one real workflow end-to-endClose 1 pilot userProof of value
4. CommissionValidate against PRD criteriaCapture testimonial + case dataReusable evidence
5. ScalePromote patterns to platform primitivesExpand into adjacent verticalsCompounding GTM

The Ecosystem Compound

The mycelium is not a metaphor. It's an economic structure. Same infrastructure, different revenue surfaces.

ProductRevenue ModelEntry PointFeeds Back
ETL Data ToolProfessional services ($15-50K) → SaaS ($500/mo)Data is messy, AI can't workClean data powers every other product
RFP Response ToolSubscription ($500-5K/mo per team)Sales teams waste 20-40 hrs per RFPApproved answers compound into library
Agent PlatformPer-receipt ($1-10) or subscriptionAgents need orchestrationReceipts prove value, build trust data
Content AmplifierIncluded in agency / SaaS tierContent needs multi-platform reachDistribution data improves targeting

Each new product costs less than 20% of the first because the horizontal layer (auth, CRM, pipeline, analytics, workflow) is shared. The vertical layer is what makes each product defensible — domain knowledge that generic tools can't replicate.

SHARED INFRASTRUCTURE (build once)
├── ETL (data readiness)
├── RFP (sales intelligence)
├── Agents (orchestration)
└── Content (distribution)


EACH PRODUCT'S DATA IMPROVES THE OTHERS

PRD Standard

PRD SectionWhat It DefinesWho Uses It
The JobWhat progress the user is makingEngineering — to understand intent
Patterns / FeaturesWhat gets builtEngineering — the RFP
Success CriteriaHow to verify it worksCommissioning — the checklist
Commissioning TableComponent-by-component statusBoth — shared progress tracker
Context LinksDomain knowledge that grounds decisionsEngineering — to understand why

Every PRD must include before build handoff:

  • Job + Trigger — what event causes demand now
  • Feature / Function / Outcome — explicit mapping
  • Business Dev — ICP, offer, channels, conversion, retention
  • Commissioning — status by component
  • Risk + Kill Signal — what would invalidate the thesis fast
  • Context — links to horizontal SaaS, vertical SaaS, industry, token layers

Starting Today

  1. Pick a PRD that pulls you — Wallet Safety or Prompt Deck if you want a challenge, a Starter from the backlog if you want to ease in
  2. Read the spec — understand the job, the patterns, the success criteria
  3. Build — follow the spec, ship something
  4. Verify — does what shipped match what was specified?
  5. Feed back — update the commissioning table, update the PRD with learnings

Not positive thinking. Positive doing. Optimism is a muscle. Building is the gym.

Context