FIN Provisional Patent And Trade Secret Filing Readiness

Date: 2026-05-13 Status: Draft for counsel / founder review Prepared from active repo review, FIN Intra_GTM planning, and existing patent planning docs.

Important: this is not legal advice. It is a technical/product invention inventory and trade-secret boundary memo to help prepare a provisional patent filing discussion with qualified patent counsel.

1. Source Materials Reviewed

Existing Patent / IP Materials

Path	Role
/Users/robert/Projects/FIN/_03_Dev/FIN_Provisional_Patent_Audit_2026-04-07.md	Latest primary patent audit; identifies 12 provisional patent candidates.
/Users/robert/Projects/FIN/_03_Dev/FIN_Provisional_Patent_Audit_2026-04-07.docx	DOCX companion to primary audit.
/Users/robert/Projects/FIN/_03_Dev/FIN_Platform/docs/Patent_Strategy.md	Older urgent patent/trademark/trade secret strategy.
/Users/robert/Projects/FIN/_03_Dev/FIN_Platform/docs/fundraising/IP_STRATEGY_ROADMAP.md	Older IP roadmap; includes patents, trademarks, trade secrets, copyright framing.

New Planning Materials That Post-Date The April Patent Audit

Path	Role
FIN Intra_GTM/	Current full-shape planning set: intranet, GTM, offers, Nathel anchor, FIN Central, federal, tiering.
docs/audits/FIN_INTRA_GTM_PLANNING_CODEBASE_GAP_AUDIT_2026-05-13.md	Current planning-to-code gap audit.
docs/design/FIN_INTRANET_HUMAN_AGENT_PROJECTION_LAYER_DESIGN_2026-05-13.md	Current human-agent projection layer design.
docs/engineering/FIN_INTRANET_HUMAN_AGENT_PROJECTION_ENGINEERING_SPEC_2026-05-13.md	Current engineering spec for intranet/projection/MCP layer.
docs/processes/FIN_INTRANET_HUMAN_AGENT_PROJECTION_EXECUTION_PROGRAM_2026-05-13.md	Current execution program.
docs/strategy/FIN_INTRA_GTM_STRATEGIC_RECOMMENDATIONS_2026-05-13.md	Current strategic recommendations.

2. Filing Pressure Assessment

The decision to wait made sense while the project shape was unstable. The shape is now clearer:

• event-sourced operational core
• Knowledge Forge / FIN Central produce intelligence
• MCP as agent contract layer
• human-agent projection intranet
• GitHub-backed authored docs plus event-projected docs
• artifact/document/event/content graph
• FIN Central multi-tenant control plane
• Nathel anchor proof loop
• tiered offers from MCP to full ERP

Recommendation: move to filing preparation now. The provisional should be broad enough to capture the newly clarified system shape, not only the older April 7 technical candidates.

3. Existing Patent Candidate Baseline

The April 7 audit identified 12 candidates:

#	Existing candidate	Keep?	Current recommendation
I	Event-Sourced Agentic ERP for Perishable Goods with Knowledge Graph Intelligence	Yes	Expand to include human-agent projection layer and FIN Central control plane as embodiments.
II	Automated Food Recall Blast Radius Modeling via Knowledge Graph Traversal	Yes	Keep as high-priority; add proof-package/audit projection embodiments.
III	Hybrid Deterministic-Agent Blueprint Execution Engine with Saga Compensation	Yes	Keep; connect to portal/editor policy gates and doc mutation workflows.
IV	Three-Tier Commodity Intelligence Retrieval with Matryoshka Embeddings	Yes	Keep; connect Knowledge Forge to Tier 0 MCP and intranet citations.
V	Agent-Mediated Blind Marketplace with Graduated Identity Disclosure	Yes, but later	Update to current verified-customer marketplace and commitment-gated price reveal, not only blind marketplace.
VI	Bidirectional Intelligence Flywheel Architecture for Supply Chain Systems	Yes	Expand with content graph, portal signals, and intranet/doc feedback loops.
VII	Session-Scoped Idempotent Anonymization Proxy	Yes	Keep; add doc/MCP/projection privacy use cases.
VIII	Federated Intelligence Network	Yes, with caution	Include high-level methods; keep exact training data, aggregation parameters, and customer signals secret.
IX	AI Vision-Managed Warehouse Operations	Yes	Keep; not central to immediate filing unless budget supports broad filing.
X	Offline-Resilient Agentic ERP with FSMA-Priority Sync	Yes	Keep; valuable operational moat.
XI	Dynamic Shelf-Life Prediction Engine	Yes	Keep; strong perishable-domain specificity.
XII	Combined Intelligence Analysis / Multi-Source Signal Fusion	Yes	Keep; add latest pricing/supply/demand ingestion lifecycle if current code supports it.

4. New Invention Candidates To Add

Candidate XIII: Human-Agent Projection Intranet For Event-Sourced Enterprise Operations

Recommended priority: P0 / include in immediate provisional.

Working title:

System and Method for Synchronized Human and Agent Projections From an Event-Sourced Operational Store

Core invention:

A tenant-scoped system that converts immutable operational events, authored documents, artifacts, correspondence, and semantic memory into synchronized human-readable and agent-readable projections. The human surface is an intranet, portal, file viewer, document graph, or proof package. The agent surface is a set of structured MCP tools over the same content graph.

Key elements to disclose:

• event store as operational truth
• GitHub/markdown authored document ledger
• system-projected documents generated from events
• artifacts and correspondence as content nodes
• content index in knowledge
• chunks/embeddings in operational_memory
• actor-specific state in actors.memory
• permissioned render for humans
• permissioned MCP query for agents
• provenance/citation metadata joining docs, artifacts, events, and source files
• role-scoped projections for internal intranet, employee portal, customer portal, marketplace, and FIN Central

Why it matters:

This is the missing layer between an agent-native ERP backend and human adoption. The same operational truth becomes readable to people and queryable by agents without duplicating systems of record.

Relationship to existing candidates:

• Could be a new standalone provisional.
• Could also be an embodiment or continuation path under Application I (event-sourced agentic ERP) and Application VI (intelligence flywheel).

Candidate XIV: Git-Backed Authored And Event-Projected Document Graph With MCP Access

Recommended priority: P0 / include in immediate provisional, either standalone or as a dependent system under Candidate XIII.

Working title:

Git-Backed Enterprise Content Graph With Event-Generated Documents And Agent Tool Access

Core invention:

A system in which human-authored markdown documents and system-generated documents coexist in a Git-backed vault. Ingest parses frontmatter, tags, wikilinks, backlinks, headings, and chunks; indexes the structure into an operational knowledge projection; emits lifecycle events; exposes graph-aware MCP tools; and triggers human renderers such as an intranet.

Key elements to disclose:

• GitHub commit webhook as authored-source trigger
• event subscription as projected-doc trigger
• frontmatter/wikilink/backlink parser
• deterministic doc IDs and source paths
• commit SHA and source event IDs as provenance
• graph neighborhood retrieval
• decision record lookup
• doc search combining structured filters and embeddings
• role-scoped rendering and MCP access
• draft/review workflow for reverse projection

Trade-secret caution:

Disclose enough parser/graph/indexing mechanics to support the invention, but do not disclose exact internal prompts, ranking weights, source priority scores, or customer-specific doc taxonomies beyond illustrative examples.

Candidate XV: FIN Central Multi-Tenant Agentic Operations Control Plane

Recommended priority: P1 / include in immediate provisional if filing a broad umbrella application; otherwise queue for follow-on provisional.

Working title:

Multi-Tenant Control Plane For Agentic ERP Deployments Across Edge, Database, Document, Communication, And Agent Runtime Resources

Core invention:

A control plane that provisions, monitors, and governs tenant-specific deployments across Cloudflare, Neon, GitHub vaults, storage, MCP keys, communications channels, portal entitlements, and premium visible-agent workstations such as Orgo. The control plane is itself event-sourced and emits readiness projections, channel health, proof package status, and audit views.

Key elements to disclose:

• tenant readiness lifecycle
• Cloudflare route/resource readiness
• Neon branch/schema readiness
• GitHub vault/readiness state
• doc ingest/render readiness
• MCP key status and usage metering
• email/SMS/voice/WhatsApp channel health
• portal entitlement state
• Orgo/visible-agent status
• Nathel proof ledger integration
• federal readiness and audit export status

Why it matters:

This is the operational system needed to scale from one anchor deployment to multi-tenant agentic ERP deployments without manual chaos.

Candidate XVI: Auditable Proof Package Generation From Operational Events, Artifacts, And Agent Summaries

Recommended priority: P1 / include as embodiment in Candidate XIII or XV.

Working title:

System and Method for Generating Auditable Business Proof Packages From Event-Sourced Operations

Core invention:

A system that composes proof packages from immutable events, artifacts, screenshots, operational metrics, agent-generated summaries, source attribution, and stakeholder-specific claims. Proof packages support customer success, compliance, federal readiness, funding narratives, and business review materials.

Key elements to disclose:

• proof package as a content node
• links to source events, artifacts, docs, screenshots, and metrics
• before/after claims with evidence IDs
• stakeholder-specific views
• customer-safe redaction policies
• agent-generated narrative with citations
• event lifecycle for proof package creation, review, publication, and export

Why it matters:

This bridges system truth and commercial proof. It is likely valuable even if not the central patent claim.

Candidate XVII: Graduated Portal-To-Agent Entitlement System

Recommended priority: P1 / include as embodiment in Candidate XIII and V.

Working title:

Graduated Trust And Entitlement System For Progressively Enabling Agentic Operations In Supply Chain Portals

Core invention:

A system for onboarding external customers, vendors, or employees through graduated entitlements:

1. portal-only read/search/track
2. portal-plus-agent ask/draft/escalate
3. agent-assisted workflow with human approval
4. bounded agent execution under deterministic policy gates

Key elements to disclose:

• role/surface/tenant/entity entitlement matrix
• customer/vendor-safe projection filters
• deterministic policy checks before agent action
• audit events for portal views and agent interactions
• upgrade path based on trust, identity, and permissions
• commitment-gated marketplace price reveal as a specialized embodiment

Why it matters:

This is a practical adoption mechanism for agentic systems in conservative industries. It is more technical when tied to permissions, event sourcing, and MCP/action gating than when described as sales strategy.

Candidate XVIII: Visible Executive Agent Workstation With Event-Sourced Audit And Human Oversight

Recommended priority: P2 / include if counsel sees enough novelty; otherwise keep as trade-secret-heavy product roadmap.

Working title:

Visible Agent Workstation For Audited Executive Operations Across Enterprise Systems

Core invention:

A premium executive agent operates through a visible cloud workstation, such as Orgo, while every observed action, tool call, credential boundary, human approval, and system result is recorded into the event store and projected into executive/operator summaries.

Key elements to disclose:

• visible agent session
• event logging of observations/actions
• human approval checkpoints
• credential boundaries
• task-specific runbooks
• executive briefing projections
• linkage to intranet docs and operational events

Trade-secret caution:

Keep actual executive workflows, customer-specific automations, prompts, and operational playbooks secret.

Candidate XIX: Dynamic Pricing Engine For Perishable Goods With Event-Sourced Operational State

Recommended priority: P0/P1. The April 7 audit covers dynamic pricing inside Application XII, but the May 13 planning makes it important enough to name explicitly. Counsel should decide whether to draft it as an independent provisional family, an independent claim set under the intelligence filing, or a major embodiment of the marketplace/projection filing.

Working title:

Dynamic Pricing Engine For Perishable Goods Using Event-Sourced Operational State, Multi-Source Market Intelligence, And Permissioned Price Projection

Core invention:

A pricing engine generates customer-specific prices or price recommendations for perishable goods by combining tenant operational events, inventory velocity, receiving and quality events, inbound arrivals, customer/account terms, external market intelligence, demand signals, supply-risk signals, shelf-life constraints, and price-elasticity estimates. The engine projects pricing differently across internal operators, customer portals, marketplace views, and agent tools, with auditability, human overrides, and optional commitment-gated price reveal.

Why this should not be buried under “data fusion”:

Combined signal fusion describes how signals are gathered and interpreted. Dynamic pricing describes a downstream decision system that converts those signals into prices, price ranges, recommendations, overrides, and customer visible projections. It should be explicitly disclosed so later claims are not limited to passive intelligence briefs.

Key elements to disclose:

• event-sourced operational inputs such as inventory, lot status, inbound arrivals, order velocity, quality exceptions, claims, and recall risk
• universal market intelligence inputs such as USDA market data, weather, import disruptions, season transitions, regional supply, and trend data
• relationship/account inputs such as customer tier, terms, history, credit, buying patterns, and customer-specific restrictions
• commodity ontology inputs such as shelf life, perishability, temperature sensitivity, pack size, grade, region, and substitution relationships
• price-elasticity and demand/supply signal estimation
• generation of recommended prices, price bands, quote actions, and operator explanations
• human override workflow with event logging and provenance
• role-scoped projection of pricing into internal views, customer portals, marketplace views, and MCP tools
• commitment-gated price reveal as a specialized marketplace embodiment
• audit trail and proof package linkage showing which events/signals supported a price or recommendation

Trade-secret caution:

Disclose the architecture, inputs, outputs, control flow, override pattern, and projection mechanics. Keep exact formulas, weights, thresholds, customer terms, elasticity calibration data, margin rules, negotiation scripts, and commercial playbooks secret unless counsel says a limited example is needed for enablement.

Candidate XX: Event-Sourced Agentic QC Evidence Fabric For Perishable Goods

Recommended priority: P0/P1. The April 7 audit covers vision-managed warehouse operations and shelf-life prediction, but the active codebase now shows a broader QC system that should be named explicitly.

Working title:

Event-Sourced Agentic Quality-Control Evidence Fabric For Perishable Goods Receiving, Inspection, Disposition, And Compliance

Core invention:

A quality-control system guides a human inspector through progressive per-lot inspection steps while agents supply commodity context, grade tolerances, photo analysis, and recommended actions. Each step is recorded as an immutable event, then folded into canonical inspection-completion evidence, hold/release gates, USDA inspection requests, claim workflows, reinspection triggers, smart notifications, proof packages, and future training data.

Why this should not be buried under “vision QC”:

Vision analysis is only one input. The stronger invention is the workflow and evidence fabric: per-lot mobile cards, dual-label provenance, event-derived next steps, deterministic compliance gates, human inspector authority, structured photo/narrative/temperature/disposition events, and downstream regulatory and financial actions.

Key elements to disclose:

• progressive per-lot inspection cards driven by event-stream state
• per-lot steps for pulp temperature, photo evidence, narrative, and disposition
• dual-label capture for vendor case labels and tenant/ERP labels
• preservation of external ERP lot numbers while adding FIN-side sub-lot annotations such as harvest split, GTIN-from-photo, vendor aliases, and FIN traceability codes
• deterministic compliance gates for FSMA / traceability data before release
• commodity-aware quality scoring using visual, temperature, size/grade, Brix, and defect signals
• three-point trailer temperature variance detection across nose/middle/tail
• privacy-gated QC vision analysis for raw images, including local/full-provider modes and approved external exceptions
• human inspector judgment as authoritative training-label source until calibration thresholds are met
• canonical quality.inspection_completed evidence packages with line-item photos, narrative, Produce Pro references, source events, and retention policy
• hold/release lifecycle and sellability gate integration
• automatic routing to inspection reports, notifications, USDA handoff, and claim workflows
• reinspection triggers from shelf-life risk, random audit, and complaint traceback

Trade-secret caution:

Disclose the event model, workflow shape, evidence packet, gate structure, and downstream actions. Keep exact scoring weights, defect thresholds, calibration counts, vision prompts, role-routing heuristics, Nathel-specific SOPs, vendor names, photo datasets, and training-label corpora secret unless counsel says a sanitized example is needed.

Candidate XXI: QC-Driven Regulatory And Commercial Escalation Pipeline

Recommended priority: P1. This can be a dependent claim family or major embodiment under Candidate XX, but it should be visible to counsel because it connects warehouse QC to dollars, claims, USDA process, and PACA compliance.

Working title:

Quality-Control Triggered Regulatory, Sellability, And Commercial Claims Pipeline For Perishable Goods

Core invention:

QC events automatically trigger a sequence of regulatory and commercial actions: place a lot on hold, block sellability, generate or prefill inspection reports, request USDA inspection, notify role-specific stakeholders, file quality claims, calculate claim value, detect repeat claim patterns, generate credit workflows, and monitor PACA deadlines or legal holds.

Key elements to disclose:

• event-driven hold placement and release with sellability gate enforcement
• automatic USDA request generation, PDF/storage/email submission, and hold label creation from QC context
• vendor/grower rejection notification generated from inspection evidence
• claim creation from QC or delivery exceptions with photo/report attachments
• claim amount calculation from condition codes, units, invoice pricing, and freight offsets
• repeat-claim pattern detection
• PACA deadline monitoring and escalation recommendation
• source-event linkage so commercial actions can be traced back to inspection evidence

Trade-secret caution:

Disclose the regulated/commercial escalation architecture and event-to-action links. Keep actual PACA/legal playbooks, negotiation strategies, vendor-specific claim rates, condition-code tuning, customer/vendor names, and settlement history secret.

5. Suggested Provisional Filing Architecture

Option A: One Broad Umbrella Provisional

Title:

Agentic Perishable Goods Operations Platform With Event-Sourced Data, Human-Agent Projections, And Multi-Tenant Control Plane

Include:

• Applications I, III, VI, VII, IX, X, XI, XII, XIII, XIV, XV, XVI, XVII, XIX, XX
• brief embodiments for Tier 0 MCP, Knowledge Forge, portals, Nathel proof, Orgo, dynamic pricing, QC escalation, marketplace, and federal readiness

Pros:

• fastest path to a filing date for the current system shape
• captures the newly clarified business/product architecture
• creates room for later continuations/divisionals

Cons:

• may be broad and less crisp
• counsel may later split into multiple applications

Option B: Four Focused Immediate Provisionals

1. Agentic ERP Core And Workflow Governance
   • I, III, VII, X, XX
2. Human-Agent Projection Intranet And Content Graph
   • XIII, XIV, XVI, XVII
3. Knowledge/Intelligence And Multi-Tenant Control Plane
   • IV, VI, XII, XV, XIX
4. Perishable QC, Vision, Shelf-Life, And Escalation
   • IX, XI, XX, XXI

Pros:

• cleaner drafting
• easier attorney review
• better continuation/divisional structure

Cons:

• more filing work
• more cost

Recommendation

If time pressure is highest, file Option A first with broad disclosure and many figures. Then have counsel split or follow with more focused provisionals within the 12-month window. If budget/time permits, Option B is cleaner.

6. Current Inclusion Priority

Must Include In Immediate Provisional

Candidate	Reason
Event-sourced agentic ERP core	Foundational architecture and already central to codebase.
Human-agent projection intranet	Newly clarified missing layer; central to future product.
Git-backed authored/system-projected document graph	Specific technical mechanism for the intranet.
MCP access over content graph	Agent-facing contract and key differentiator.
FIN Central control plane	Needed for multi-tenant deployment and GTM shape.
Proof packages	Bridges technical auditability and commercial proof.
Permissioned role/surface/entity projections	Critical for customer/employee/vendor portals.
Knowledge Forge / three-plane intelligence	Core data moat and Tier 0 MCP product.
Dynamic pricing engine	Distinct downstream decision system that converts event, market, demand, supply, customer, and perishability signals into prices, recommendations, overrides, and permissioned projections.
Event-sourced QC evidence fabric	Central Nathel proof path; combines human inspection, agent guidance, mobile cards, event-sourced evidence, compliance gates, vision, holds, reports, and downstream actions.
Deterministic-agent workflow gates	Protects agent execution in regulated operations.
Anonymization/privacy boundary	Critical for external LLM and customer data handling.

Should Include If Space/Time Allows

Candidate	Reason
Offline FSMA-priority sync	Strong operational/regulatory specificity.
Shelf-life prediction	Strong perishable-domain specificity and current code evidence.
Combined signal fusion	Important upstream intelligence layer for pricing/supply/demand decisions.
Federated intelligence	Strategic moat, but exact mechanics should be carefully disclosed.
Vision/RFID/sensor fusion	Strong but less central to immediate intranet/GTM shape.
Orgo visible executive agent	Useful Tier 3 embodiment, but may be better after contract/spec.
Marketplace commitment-gated price reveal	Include as dynamic pricing / marketplace embodiment; avoid over-centering before build.
QC-driven USDA/PACA/claims escalation	Include as an embodiment or dependent claim family under the QC evidence fabric; commercially important but may not need to stand alone.

Do Not Center As Patent Claims

Topic	Reason
Pricing tiers	Business model, not core technical invention.
Founder-led sales	GTM execution, not patent subject matter.
Nathel as named anchor	Commercial strategy and confidential relationship.
Productized consulting	Business/process idea; keep as strategy.
Generic “dynamic pricing”	Prior-art-heavy unless tied to event-sourced operational state, perishable constraints, customer-specific terms, market intelligence, permissioned projections, audit, and human override.
Generic “quality inspection app”	Prior-art-heavy unless tied to event-sourced per-lot progression, human-agent evidence capture, FSMA/USDA/PACA actions, sellability gates, and training feedback.
Generic use of GitHub, Markdown, Obsidian, Quartz	Prior-art-heavy unless tied to event/MCP/projection mechanics.
Generic “AI intranet”	Too broad; must be tied to concrete technical mechanism.

7. Trade Secret Inventory

Trade secrets should include the valuable details that are not necessary to enable the patent filing and are not easily reverse-engineered from the product.

7.1 Data And Customer Relationship Secrets

Keep secret:

• Nathel operational history and customer/vendor data
• customer pricing, margins, cost basis, terms, and rebates
• vendor score details and quality histories
• customer segmentation and buying patterns
• enterprise correspondence archives
• executive personal context, calendar/email-derived memory, and preferences

Patent disclosure boundary:

• disclose categories of data and how they flow through the system
• do not disclose real customer names, actual pricing, actual customer lists, or exact commercial performance metrics unless counsel requires sanitized examples

7.2 Knowledge Forge And Data Quality Secrets

Keep secret:

• source weighting and freshness formulas
• extraction prompt wording
• curation heuristics
• conflict resolution thresholds
• source priority rankings
• crawler scheduling heuristics
• exact corpus quality scores
• curated commodity profiles not publicly released

Patent disclosure boundary:

• disclose the existence of multi-source ingestion, structured commodity profiles, embeddings, citations, and quality stages
• keep exact weights, prompts, source schedules, and tuning parameters private

7.3 Agent Runtime And Prompting Secrets

Keep secret:

• system prompts
• role/persona prompt packs
• tool-selection heuristics
• eval scenario libraries
• scoring rubrics and grade thresholds
• production fallback policies
• model routing thresholds
• prompt-injection defenses beyond high-level descriptions

Patent disclosure boundary:

• disclose agent/tool orchestration, deterministic gates, MCP interfaces, and anonymization method
• keep exact prompts, heuristics, and tuning private

7.4 Pricing, Marketplace, And GTM Secrets

Keep secret:

• dynamic pricing formulas
• customer-specific price reveal thresholds
• commitment-gating thresholds
• marketplace matching/reputation weights
• arbitrage detection logic
• sales scripts and negotiation playbooks
• Nathel funding strategy and commercial asks

Patent disclosure boundary:

• disclose graduated price reveal and marketplace projection mechanics
• keep exact pricing formulas, commercial terms, and playbooks secret

7.5 FIN Central Operations Secrets

Keep secret:

• deployment runbooks that expose operational vulnerabilities
• tenant provisioning scripts with real secrets/config names
• internal readiness thresholds
• incident response playbooks
• Cloudflare/Neon/Orgo operational details not needed for enablement

Patent disclosure boundary:

• disclose the control-plane architecture and lifecycle states
• keep exact credentials, scripts, thresholds, and operational shortcuts private

7.6 Orgo / Executive Agent Secrets

Keep secret:

• specific executive workflows
• watched portal credentials
• vendor/customer interaction scripts
• human approval heuristics
• personal memory and preference structures

Patent disclosure boundary:

• disclose visible agent workstation with event-sourced audit and approval gates
• keep actual customer/executive implementations private

8. Drawings / Figures Needed

Recommended provisional figures:

1. Four-table event-sourced architecture with five FIN components.
2. Three data planes: private workspace, relationship data, universal intelligence.
3. Human-agent projection layer: events + docs + artifacts + opmem to human surfaces and MCP.
4. GitHub authored vault and event-projected document ingest pipeline.
5. Content graph showing docs, artifacts, events, actors, customers, vendors, lots, commodities, proof packages.
6. Permissioned projection matrix across internal intranet, employee portal, customer portal, marketplace, FIN Central, public MCP.
7. MCP tool access flow with citations and tenant scoping.
8. FIN Central multi-tenant control plane across Cloudflare, Neon, GitHub, storage, comms, MCP keys, Orgo.
9. Nathel proof package lifecycle from event/artifact capture to stakeholder report.
10. Graduated portal-to-agent entitlement ladder.
11. Deterministic-agent blueprint workflow with safety gates and compensation.
12. Knowledge Forge to FIN Central KB to Tier 0 MCP flow.
13. Anonymization proxy before external LLM inference.
14. Dynamic pricing engine: event store + relationship data + market intelligence + perishability ontology to internal/customer/marketplace projections.
15. QC evidence fabric: per-lot mobile inspection card loop, photo/narrative/temp/disposition events, canonical inspection packet, hold/release gate, report/proof package.
16. QC escalation pipeline: QC event to USDA request, sellability block, vendor/grower notification, quality claim, PACA deadline, credit workflow.
17. Optional: shelf-life / signal fusion / vision/RFID embodiment diagrams.

9. Counsel Packet Checklist

Prepare for counsel:

• April 7 patent audit
• May 13 planning/code gap audit
• May 13 design/spec/execution docs
• May 13 strategic recommendations
• this filing readiness memo
• AGENTS.md architecture law
• artifact catalog integration doc
• code evidence list for active files
• diagram set
• invention disclosure forms for Candidates XIII-XXI
• trade secret schedule
• NDA/IP assignment status for contributors
• list of any public disclosures, demos, investor pitches, or customer demos
• list of planned public announcements/content to hold until filing

10. Filing Risk Notes To Confirm With Counsel

Confirm:

• Whether to file one broad provisional or multiple focused provisionals.
• Whether marketplace and Orgo concepts are sufficiently enabled now or should be follow-on filings.
• Whether the dynamic pricing engine should be claimed independently or as a major embodiment of the multi-source signal fusion / marketplace filings.
• Whether a sanitized pricing example is needed for enablement while preserving formulas, weights, thresholds, and customer terms as trade secrets.
• Whether the QC evidence fabric should be claimed independently from the broader event-sourced ERP and vision-managed warehouse filings.
• Whether the USDA/PACA/claims escalation path is a dependent embodiment or a separate regulatory/commercial workflow claim family.
• Whether any public/private demos have started a disclosure clock.
• Whether foreign filing rights are at risk from any prior disclosure.
• Whether GTM-only content should be removed from the patent spec and retained as trade secret.
• Whether “MCP” should be described generically as a structured agent tool protocol because MCP itself is an open standard.
• Whether claim language should avoid “exactly four tables” as a limiting requirement or present it as one embodiment.

11. Immediate Action Plan

1. Freeze public disclosure of the new intranet/projection/MCP/FIN Central/dynamic pricing/QC evidence concepts until provisional filing.
2. Decide filing architecture: one broad umbrella or four focused filings.
3. Create figures 1-16 above, with figure 17 if counsel wants optional embodiments.
4. Draft invention disclosures for Candidates XIII-XXI.
5. Attach April 7 candidates I-XII as baseline.
6. Mark trade secret schedule and remove specific weights/prompts/customer data from filing draft.
7. Review with counsel.
8. File provisional.
9. Calendar 12-month nonprovisional/PCT deadline immediately after filing.

12. Bottom Line

The April 7 patent audit is still valuable, but it is incomplete relative to the current business shape. The May 13 planning adds a new central invention family: the human-agent projection intranet, Git-backed/event-projected document graph, MCP-accessible content graph, proof package system, portal entitlement ladder, FIN Central multi-tenant control plane, explicit dynamic pricing engine, and event-sourced QC evidence fabric.

Those ideas should be added to the provisional disclosure before filing. The exact prompts, data, weights, customer records, pricing formulas, deployment details, and operator playbooks should be handled as trade secrets.