FIN Central

FIN Platform North Star Roadmap

14 min read

FIN Platform North Star Roadmap

0. Revision Summary

First cut. The meta-roadmap. Names the eleven pillars that compose the full FIN agentic OS goal, marks the status of each, links to the canonical doc(s) per pillar, and lays out the 6-month and 18-month horizons. This is the doc to read once a month to know exactly where the system stands and what is still missing.

1. Purpose

FIN’s stated goal is an agentic operating system for a high-ops-tempo, high-data-volume produce wholesaler in NYC. That goal is a system-of-systems. No single roadmap captures it. This doc is the index of indexes — the canonical view of the eleven pillars that compose the platform and the canonical roadmap per pillar.

What this doc is:

  • The answer to “do I have a complete blueprint?”
  • The map between the strategic goal and the canonical body of work
  • The dependency graph between pillars
  • The monthly review surface for platform-wide status

What this doc is not:

  • A replacement for any pillar’s canonical roadmap
  • An execution-level program (those live in docs/programs/)
  • A status report (those are generated from the gradeboard and the doc index)

2. The Eleven Pillars

The pillars and their canonical homes.

Pillar 1 — Architectural Substrate

Events, CQRS projections, MD projection layer, agentic harness, four-table law. The bedrock that everything else sits on.

Pillar 2 — Operational Data Corpus

Every signal from every operational system, bound to PO spine. PP, TURVO, Samsara, comms, QC, sales, inventory.

  • Status: Roadmap landed (2026-05-19)
  • Canonical doc: Operational Corpus Ingestion Roadmap
  • 6-month target: Nathel corpus complete — PP + TURVO + Samsara + comms + QC flowing in continuously; ~15 new projections (migrations 049-063) live; aggregate intelligence projections (vendor/carrier/customer scorecards, sight-picture) rendering daily.
  • 18-month target: Same pipeline shape applies to ≥ 3 tenants; cross-tenant aggregate projections feed Pillar 10.
  • Gates: Depends on Pillar 1 (substrate). Gates Pillar 5 (no learning without corpus) and Pillar 9 (no curation without ingestion).

Pillar 3 — World Knowledge Corpus (Taproot)

The produce-industry KB itself — commodity reference data, grower/vendor atlas, USDA AMS Market News, FSMA reference, organic certification standards, seasonality, market intelligence. The substance behind “knows produce.”

  • Status: Roadmap landed (2026-05-19)
  • Canonical doc: Taproot Knowledge Curation Roadmap
  • 6-month target: Taproot central canon at v1 — ≥ 20 commodity profiles, top-100 NYC-area vendors and growers, AMS Market News ingest live, organic/FSMA reference docs canonical, seasonality calendars complete.
  • 18-month target: Taproot central canon at v2 with ≥ 100 commodity profiles, full national grower atlas, federal reference layer (FDA/USDA classification), Harvest Directory public surface populated.
  • Gates: Depends on Pillar 1 (substrate) and Pillar 9 (curation discipline). Gates Pillar 5 (verifiers need authoritative reference data) and Pillar 10 (Tier 0 public surface depends on this).

Pillar 4 — Interface Layer

QC system, comms system, intranet (FIN + Nathel), voice (Telnyx), mobile PWA, agentic console. The surfaces operators and partners actually touch.

  • Status: Partial — major surfaces have canonical docs, but no integrated multi-surface roadmap exists
  • Canonical docs:
  • 6-month target: QC and comms live at Nathel; intranet (FIN + Nathel) live; voice via Telnyx live; PWA shipped to first inspector cohort and operator cohort. Integrated interface roadmap doc landed (gap to close in next 60 days).
  • 18-month target: Full operator console, customer-facing surfaces (Harvest Directory + customer portals), partner surfaces (vendor scorecards external-facing), federal surfaces (compliance reporting).
  • Gates: Depends on Pillar 1 (substrate) and Pillar 8 (adoption). Gates Pillar 11 (no business OS without surfaces).
  • Gap: Need an integrated interface roadmap doc — docs/design/FIN_INTERFACE_ROADMAP_2026-Q3.md. Target: 60 days.

Pillar 5 — Closed-Loop Learning

Verifiers, procedure projections, self-critique loop, case memory, counterfactual replay, confidence calibration, skill packaging. The self-improving layer.

  • Status: Roadmap landed (2026-05-19)
  • Canonical doc: Self-Improving System Roadmap
  • 6-month target: Phase A + Phase B complete — deterministic verifier suite running inline on every QC, comms, intranet agent response; first system-proposed skill edit merged; counterfactual replay v0 demonstrated.
  • 18-month target: Phases C + D complete — case memory live, confidence calibration live, learned verifier in production, distilled domain model evaluated.
  • Gates: Depends on Pillar 1 (substrate), Pillar 2 (corpus for evidence), Pillar 3 (reference data for verifiers). Gates the credibility of every other agent surface.

Pillar 6 — Reliability and Ops at High Tempo

Latency budgets, throughput targets, queue depth budgets, observability stack, SLO/SLI framework, incident response, DR, on-call. The discipline of running this thing inside an active NYC produce wholesaler.

  • Status: Roadmap landed (2026-05-19)
  • Canonical doc: Reliability and Ops Roadmap
  • 6-month target: Production SLOs defined and met per surface; observability stack deployed (logs, metrics, traces, gradeboard, harness telemetry); on-call rotation established; first DR drill complete.
  • 18-month target: 99.9% availability on Tier 2 surfaces; multi-region failover ready; mature incident playbooks across all surfaces.
  • Gates: Depends on Pillar 1 (substrate). Gates Pillar 11 (no federal pipeline without reliability evidence) and Pillar 10 (no scale-out without operational maturity).

Pillar 7 — Security, Compliance, Governance

Anonymization, retention, audit trail, federal posture, FSMA 204 traceability, SOC 2, GDPR/CCPA, government data classification. The trust layer.

  • Status: Partial — Anonymization Policy v1 landed; federal pipeline doc exists; provisional patent plan in motion. No integrated compliance roadmap.
  • Canonical docs:
  • 6-month target: Compliance roadmap doc landed (gap to close in Q3). FSMA 204 KDE traceability demonstrable end-to-end. SOC 2 Type I evidence collection started. Provisional patents filed.
  • 18-month target: SOC 2 Type II achieved. FSMA 204 third-party audited. SDVOSB pursued if eligible. Federal pilot live.
  • Gates: Anonymization Policy gates Wave 3 (any Plane 3 publication). Compliance roadmap gates Pillar 10 (no public/federal surfaces without).
  • Gap: Need integrated Compliance Roadmap doc — docs/policies/FIN_COMPLIANCE_ROADMAP_2026-Q3.md. Target: 90 days.

Pillar 8 — Human Adoption

Operator onboarding, change management, training curriculum, support model, escalation paths, operator velocity tracking. The discipline that gets humans actually using the system.

  • Status: Partial — agentic onboarding program for Nathel cohort exists; UX design suite exists. No integrated adoption roadmap.
  • Canonical docs:
  • 6-month target: Nathel cohort fully onboarded (3 inspectors + Alex + Josh + Jake actively using). Adoption roadmap doc landed. Operator velocity metrics defined and tracked in gradeboard.
  • 18-month target: Multi-tenant adoption playbook proven; ≥ 3 tenants onboarded under standard playbook; operator velocity benchmarks established for each Tier.
  • Gates: Depends on Pillar 4 (interfaces). Gates Pillar 10 (no scale-out without proven adoption motion).
  • Gap: Need Adoption Roadmap doc — docs/programs/FIN_ADOPTION_ROADMAP_2026-Q3.md. Target: 90 days.

Pillar 9 — Knowledge Integrity (Curation Discipline)

The discipline of maintaining a world-class produce data set over time. Master data quality, resolver discipline, deduplication, drift detection, ground-truth selection, ongoing curation of Taproot. The “world-class data set” requirement Rob named explicitly.

  • Status: Roadmap landed (2026-05-19) — combined with Pillar 3 in Taproot Knowledge Curation Roadmap
  • Canonical doc: Taproot Knowledge Curation Roadmap (sections §4-§7 specifically address curation discipline)
  • 6-month target: Resolver v2 production-grade; drift monitoring live for AMS Market News + organic/FSMA reference sources; curation workflow with proposed/approved/published states for every Taproot entry.
  • 18-month target: Curation discipline mature enough to support multi-tenant overlays and federated knowledge contributions.
  • Gates: Depends on Pillar 3 (the corpus to curate). Gates Pillars 5, 10, 11.

Pillar 10 — Multi-Tenant Scale-Out

Going from Nathel to N tenants. Federated intelligence. Marketplace. FINN consumer surface. Harvest Directory public. Tier 0 public MCP. Federal sovereign deployment.

  • Status: Partial — marketplace planning doc exists; Tier Architecture exists; Demarcation Plan exists. No integrated scale-out roadmap.
  • Canonical docs:
  • 6-month target: Nathel pilot proven. Second tenant identified, demarcation infrastructure validated against second deployment. Scale-out roadmap doc landed.
  • 18-month target: ≥ 3 tenants on Tier 2+. Tier 0 public MCP live with rate-limited surface. Federated intelligence demonstrated with ≥ 3 contributing tenants.
  • Gates: Depends on Pillars 1, 2, 5, 6, 7, 8. The integration test of the entire platform.
  • Gap: Need Scale-Out Roadmap doc — docs/programs/FIN_SCALE_OUT_ROADMAP_2026-Q4.md. Target: 120 days.

Pillar 11 — Business Operating System

Federal pipeline motion, partner motion, sales motion, finance/accounting, legal/IP. The company-side discipline that makes the platform commercially viable.

  • Status: Partial — federal pipeline doc; founder bio with VOSB credentials; provisional patent plan in motion; Burnout Protection doc. No integrated business OS roadmap.
  • Canonical docs:
  • 6-month target: Provisional patents filed. SAM.gov registered fully. First federal touchpoint scheduled. Business OS roadmap doc landed.
  • 18-month target: Federal pilot signed; counsel-led non-provisional patent applications filed; revenue model proven; partnership pipeline mature.
  • Gates: Depends on Pillars 6, 7 (federal demands reliability + compliance). Gates eventual scale.
  • Gap: Need integrated Business OS Roadmap doc — docs/programs/FIN_BUSINESS_OS_ROADMAP_2026-Q4.md. Target: 150 days.

3. Pillar Status Matrix

#PillarStatusCanonical RoadmapDoc Path
1Architectural SubstrateLockedYes (multiple)engineering/, architecture/
2Operational Data CorpusRoadmap landedYesarchitecture/FIN_OPERATIONAL_CORPUS_INGESTION_ROADMAP_2026-05-19.md
3World Knowledge CorpusRoadmap landedYesarchitecture/FIN_TAPROOT_KNOWLEDGE_CURATION_ROADMAP_2026-05-19.md
4Interface LayerPartial — gapNo (interfaces docs exist; no integrated roadmap)design/ (multiple)
5Closed-Loop LearningRoadmap landedYesarchitecture/FIN_SELF_IMPROVING_SYSTEM_ROADMAP_2026-05-19.md
6Reliability and OpsRoadmap landedYesarchitecture/FIN_RELIABILITY_AND_OPS_ROADMAP_2026-05-19.md
7Security, Compliance, GovernancePartial — gapNo (Anonymization Policy v1 exists; broader roadmap gap)policies/
8Human AdoptionPartial — gapNo (onboarding program exists; broader roadmap gap)programs/, design/
9Knowledge Integrity (Curation)Roadmap landedYes (combined with Pillar 3)architecture/FIN_TAPROOT_KNOWLEDGE_CURATION_ROADMAP_2026-05-19.md
10Multi-Tenant Scale-OutPartial — gapNo (component docs exist; integrated gap)product/, architecture/
11Business Operating SystemPartial — gapNo (component docs exist; integrated gap)FIN Intra_GTM/, reference/

Pillars with full canonical roadmaps: 6 of 11 (1, 2, 3, 5, 6, 9). Pillars partial with named gap: 5 of 11 (4, 7, 8, 10, 11).

4. Dependency Graph

                          ┌──────────────────────────┐
                          │  Pillar 1 — Substrate    │
                          │  (foundational, locked)  │
                          └──────────────┬───────────┘
                                         │
                ┌────────────────────────┼────────────────────────┐
                ▼                        ▼                        ▼
   ┌─────────────────────┐   ┌────────────────────┐   ┌────────────────────┐
   │ Pillar 2 — Op Corpus│   │ Pillar 3 — Taproot │   │ Pillar 4 — Interfaces│
   └──────────┬──────────┘   └─────────┬──────────┘   └──────────┬─────────┘
              │                        │                         │
              └────────┬───────────────┼─────────────────────────┘
                       ▼               ▼
            ┌────────────────────────────────┐
            │ Pillar 9 — Knowledge Integrity │
            │ (curation discipline)          │
            └─────────────┬──────────────────┘
                          │
                          ▼
            ┌────────────────────────────────┐
            │ Pillar 5 — Closed-Loop Learning│
            │ (verifiers, self-critique)     │
            └─────────────┬──────────────────┘
                          │
                          ▼
            ┌────────────────────────────────┐
            │ Pillar 6 — Reliability & Ops   │
            └─────────────┬──────────────────┘
                          │
                          ▼
            ┌────────────────────────────────┐
            │ Pillar 7 — Sec/Compliance/Gov  │
            └─────────────┬──────────────────┘
                          │
                          ▼
            ┌────────────────────────────────┐
            │ Pillar 8 — Human Adoption      │
            └─────────────┬──────────────────┘
                          │
                          ▼
            ┌────────────────────────────────┐
            │ Pillar 10 — Multi-Tenant Scale │
            └─────────────┬──────────────────┘
                          │
                          ▼
            ┌────────────────────────────────┐
            │ Pillar 11 — Business OS        │
            │ (federal/commercial)           │
            └────────────────────────────────┘

Reading: substrate gates everything; corpus + taproot + interfaces are the three first-level branches off substrate; curation discipline gates the learning loop; reliability gates compliance; compliance gates scale-out and business OS. Pillars cannot ship out of dependency order.

5. Six-Month Horizon (End of 2026, Nov-Dec)

The horizon to which active platform investment is directed.

  • Pillar 1: Substrate v0.3 amendments landed (procedure projections, doc_kind extensions, verifier hooks); harness v0.2.
  • Pillar 2: Nathel corpus complete; ~15 new projections live; aggregate intelligence projections daily.
  • Pillar 3: Taproot canon v1 — ≥ 20 commodity profiles, top-100 NYC vendors/growers, AMS Market News + FSMA + organic reference layers.
  • Pillar 4: Live surfaces at Nathel; integrated interface roadmap landed.
  • Pillar 5: Self-improving Phase A + Phase B complete; deterministic verifier suite inline.
  • Pillar 6: Production SLOs defined and met; observability stack deployed; on-call rotation established.
  • Pillar 7: Compliance roadmap landed; provisional patents filed; FSMA 204 demonstrable.
  • Pillar 8: Nathel cohort fully onboarded; adoption roadmap landed.
  • Pillar 9: Resolver v2 production; drift monitoring live; curation workflow with proposed/approved/published.
  • Pillar 10: Second tenant identified; scale-out roadmap landed.
  • Pillar 11: SAM.gov registered; first federal touchpoint scheduled; business OS roadmap landed.

By end of 2026, all eleven pillars have canonical roadmaps. Five gap-pillars (4, 7, 8, 10, 11) close within 60-150 days.

6. Eighteen-Month Horizon (End of 2027, Nov-Dec)

The horizon to which strategic platform investment is directed.

  • Pillar 1: Substrate stable; quarterly amendment cadence.
  • Pillar 2: ≥ 3 tenant corpora flowing.
  • Pillar 3: Taproot canon v2 — ≥ 100 commodity profiles, full national grower atlas, federal reference layer, Harvest Directory public surface populated.
  • Pillar 4: Full operator console + customer-facing surfaces + partner surfaces + federal surfaces.
  • Pillar 5: Phases C + D complete — case memory live, confidence calibration live, learned verifier in production, distilled model evaluated.
  • Pillar 6: 99.9% availability on Tier 2 surfaces; multi-region failover ready.
  • Pillar 7: SOC 2 Type II achieved; FSMA 204 third-party audited; federal pilot live; SDVOSB if eligible.
  • Pillar 8: Multi-tenant adoption playbook proven; ≥ 3 tenants onboarded.
  • Pillar 9: Curation discipline mature; federated knowledge contributions live.
  • Pillar 10: ≥ 3 tenants on Tier 2+; Tier 0 public MCP live; federated intelligence demonstrated.
  • Pillar 11: Federal pilot signed; non-provisional patents filed; revenue model proven; partnership pipeline mature.

7. Quarterly Review Discipline

Once per quarter, this doc is the agenda:

  1. Walk pillars 1-11 in order
  2. Confirm canonical doc is current for each
  3. Update Status column in §3 matrix
  4. Re-confirm 6-month and 18-month targets — promote, adjust, or retire
  5. Re-evaluate dependency graph — flag any inversion (something blocking a thing it shouldn’t)
  6. Identify the next quarter’s headline pillar (the one to invest most in)
  7. Identify the next quarter’s at-risk pillar (the one falling behind)

The quarterly review produces a one-page status memo posted to the FIN Central intranet under /decisions/quarterly-platform-review-YYYY-QN.md.

8. How to Use This Doc

  • Monthly: scan §3 matrix, confirm nothing has silently shifted from Locked or Landed back to Partial.
  • Quarterly: full §7 review.
  • When adding a new initiative: map it to one of the eleven pillars before starting. If it doesn’t fit any pillar, that is a signal — either the initiative is off-strategy, or there’s a missing pillar that needs to be added to this doc.
  • When making a major architectural decision: check the dependency graph; you cannot ship a higher pillar without its dependencies in place.
  • When onboarding a new contributor: this is the second doc they read, after AGENTS.md. It is the map.

9. The Punchline

Eleven pillars. Six with canonical roadmaps as of 2026-05-19. Five gap-pillars with named owners (rob) and concrete close-by dates (60-150 days). Dependency graph respects the architectural law: substrate first, corpora second, curation third, learning fourth, reliability fifth, compliance sixth, adoption seventh, scale eighth, business ninth.

This doc is the answer to “is the blueprint complete?” Yes — every layer of the agentic OS is either documented or named as a gap with a target date.

10. Cross-Doc Index

All canonical pillar roadmaps and supporting docs:

PillarCanonical Doc
1Projection Spec v0.2, Harness Spec v0.1, MCP Surface, API Layer Spec, Demarcation Plan, Projection Catalog
2Operational Corpus Ingestion Roadmap
3 + 9Taproot Knowledge Curation Roadmap
4Multiple design docs; integrated roadmap gap, target 60 days
5Self-Improving System Roadmap
6Reliability and Ops Roadmap
7Anonymization Policy v1; integrated compliance roadmap gap, target 90 days
8Agentic Onboarding Program, UX Design Suite; integrated adoption roadmap gap, target 90 days
10Tier Architecture, Demarcation Plan, Data Signal Offerings; integrated scale-out roadmap gap, target 120 days
11Federal Pipeline, Founder Bio, Provisional Patent Plan; integrated business OS roadmap gap, target 150 days

11. Audit Annotation - 2026-05-20

Roadmap audit tracker: FIN Agentic Harness and Knowledge Roadmap Audit Tracker.

Verification status: the pillar map is coherent and aligns with the four-table/event-projection architecture, but “canonical” should be read as strategic doc status until the corrected roadmap family and tracker are synced into the FIN Central intranet repo. The audit found that several May 19 architecture roadmaps were not present in fin-platform/fin-central-intranet before this correction pass.

Execution note: the next execution program should sequence deterministic verifier/measurement work before optimization work, and should use the corrected Produce Pro report mapping from the Operational Corpus roadmap annotation.

Graph View

Backlinks