ExplainersHumanoid robots

How does DEPLOY track partnership lifecycle state?

DEPLOY tracks partnership lifecycle state as a four-state framework operating at relationship-graph granularity: announced (verified-from-press-release; not-yet-active) → active (verified-deployed or verified-shipped; current_status=true) → dissolved (terminated; current_status=false with end_date populated) → unverified-current-state (announced-but-no-update-since; cap-flag honest-absence). Per Agent A's Arc C substrate, 18 partnerships + 32 parties + 14 external counterparties (via XOR pattern) + NVIDIA 4-counterparty multi-party-partnership node verified at primary-source-anchored depth. The canonical lifecycle worked example: Figure × OpenAI announced 2024 → dissolved February 2025 (status=dissolved + endDate populated). The external-name XOR pattern (partnership_parties.company_id when counterparty is tracked entity; external_name when counterparty is genuinely not-in-registry; OpenAI + Uber + Nissan + Microsoft as canonical worked examples) operates as verification-posture discipline at relationship-graph granularity. Cap-flag-as-trust-signal operates recursively on partnership framing.

18 partnerships

Arc C verified-base substrate at primary-source-anchored depth

32 parties

Partnership_parties relationship-graph records

14 external

External counterparties via XOR pattern (OpenAI + Uber + Nissan + Microsoft)

NVIDIA 4-counterparty

GR00T × 1X + Figure + Agility + Boston Dynamics multi-party node

Figure × OpenAI Feb 2025

Canonical lifecycle dissolution precedent; status=dissolved + endDate populated

Mid-2026

Snapshot date

Why partnership lifecycle is editorially central to the framework

Partnership lifecycle state sits at the intersection of verification posture and relationship-graph granularity. The verified-vs-claimed framework operates uniformly across entity records, deployment records, incident records, and now relationship records. The discipline that distinguishes DEPLOY's partnership framework from the broader announcement-discourse cohort: a partnership announcement is not a partnership in active state, and a partnership in active state is not a partnership-of-record at primary-source-anchored verification depth. Each transition between lifecycle states corresponds to a distinct verification posture; the framework reads partnership records at the lifecycle state where each fact actually resolves.

The framework operates at canonical-methodology depth, not at marketing-announcement depth. The product differentiator: DEPLOY's framework treats partnership lifecycle the same way it treats every other verified-vs-claimed claim. Honest "announced but lifecycle-not-verified-since" beats fabricated "active partnership" framing.

The four-state partnership lifecycle framework

Per Agent A's Arc C substrate (18 partnerships + 32 parties + 14 external via XOR pattern), the partnership lifecycle framework operates at four canonical states. Each state has a distinct verification posture, a distinct primary-source anchor, and a distinct cap-flag honest-absence layer.

State 1: announced. The partnership was disclosed via press-release or company-statement; counterparty identification verified; partnership scope verified at announcement-tier depth; current operational status not yet verified at deployment-tier depth. The framework reads announced state as verified-at-announcement-depth + cap-flag-honest-absence-at-deployment-tier. Press-release reading is at reputable-press source-quality-tier depth per the 9-tier source-quality rubric; not at primary-source-anchored verification depth.

State 2: active. The partnership transitioned from announced to verified-deployed or verified-shipped. Active state requires either a deployment record (Figure 02 at BMW Spartanburg 30,000 vehicles; Apptronik Apollo at Mercedes-Benz pilot facilities; Agility Digit at GXO Flowery Branch 100,000 totes) or a shipped-product record (Wayve × Stellantis STLA AutoDrive commercial integration). Active state reads at primary-source-anchored verification depth; the partnership record carries current_status=true; current-state cap-flag operates only at sub-claim depth (specific operational scale + throughput + economic terms).

State 3: dissolved. The partnership terminated. Per Agent A's Arc C substrate, dissolved state requires current_status=false + end_date populated. The canonical worked example: Figure × OpenAI announced 2024 → dissolved February 2025 (status=dissolved + endDate populated; verified at primary-source company-statement depth). Dissolved state operates at verified-termination depth; the framework reads termination as substantive editorial state change, not as soft-decay.

State 4: unverified-current-state. The partnership was announced but operational state has not been re-verified since announcement. The framework reads unverified-current-state as honest-absence cap-flag at lifecycle depth. The 1X × OpenAI partnership operates at this state per Agent A precedent (announced + reported-not-verified with explicit lifecycle-verify flag per Figure × OpenAI precedent). Cap-flag-as-trust-signal operates at lifecycle depth: the absence of post-announcement primary-source verification is the editorial truth, not an inference of active or dissolved state.

Figure × OpenAI dissolution as canonical lifecycle precedent

The Figure × OpenAI partnership operates as the canonical lifecycle precedent across the cohort. The announcement (2024) verified Figure receiving custom AI models from OpenAI for humanoid platform integration; the partnership reached active state at announcement-tier verification. February 2025 verified dissolution per primary-source company-statement; the partnership transitioned to dissolved state with end_date populated.

The precedent establishes the discipline at three layers:

  • State-transition primary-source anchoring: each lifecycle state transition requires primary-source verification at announcement-tier or termination-tier depth. The framework does not infer dissolution from announcement-decay or active state from time-since-announcement.
  • End-date specificity: dissolved state carries specific end_date verified at primary-source depth. The framework reads end_date as primary-source-anchored at month-precision minimum; year-precision operates at honest-absence cap-flag.
  • Lifecycle-verify flag for unverified-current-state: when a partnership has been announced but operational state has not been re-verified since announcement, the framework reads the partnership at honest-absence cap-flag with explicit lifecycle-verify flag. 1X × OpenAI operates at this posture per Figure × OpenAI precedent.

The external-name XOR pattern as verification-posture discipline

The external-name XOR pattern operates at relationship-graph granularity. Per Agent A's Arc C substrate, partnership_parties carries either company_id (when counterparty is a tracked entity in the registry) or external_name (when counterparty is genuinely not-in-registry). The XOR CHECK constraint at the schema layer enforces correctness: exactly one of the two fields is populated per party record.

The pattern operates at three editorial layers:

  • Tracked-entity primary-source-anchoring: when counterparty is tracked, company_id carries primary-source verification at entity-record depth. Partnership records inherit the verified-entity-record verification posture for that counterparty.
  • External-name honest-absence first-class: when counterparty is genuinely not-in-registry (the canonical worked examples include OpenAI, Uber, Nissan, Microsoft), external_name carries the verification posture at honest-absence tier. The framework reads not-in-registry as the editorial truth, not as inference-of-existence at lower verification depth.
  • Schema-layer XOR CHECK as discipline-enforcement infrastructure: structural XOR CHECK at the database layer enforces correctness. The framework treats discipline-enforcement infrastructure as load-bearing; the alternative (convention-only) operates at substantially lower verification depth at scale.

For the broader discipline-enforcement infrastructure pattern, see DEPLOY's editorial process (editorial cleanliness + type-safety gates + per-page commit cadence operate at the same discipline-enforcement infrastructure tier).

Cross-cohort partnership pattern taxonomy

The 18-partnership substrate surfaces a structurally-meaningful cross-cohort pattern taxonomy. The framework reads four canonical pattern classes:

Humanoid manufacturing partnerships. The cohort with the deepest partnership-record substrate. Verified worked examples: BMW × Figure (Figure 02 at BMW Spartanburg deployment; 30,000 BMW X3 vehicles + 1,250 hours runtime); Mercedes × Apptronik (Apptronik Apollo Mercedes deployment; dual-facility European pilots); Hyundai × Boston Dynamics (Atlas Hyundai Metaplant deployment; corporate-parent relationship per maker-facility rule); Jabil × Apptronik; Foxconn × NVIDIA. The framework reads humanoid manufacturing partnerships at end-product OEM acceptance verification depth where the deployment substrate supports (Figure 02 at BMW Spartanburg as canonical).

Humanoid logistics deployment partnerships. Verified worked examples: GXO × Apptronik (Apollo enterprise pilots); GXO × Agility (Digit at GXO Flowery Branch 100,000 totes scaled-throughput); Catalyst Brands × Figure (Figure 03 at Catalyst Brands Reno deployment). The framework reads humanoid logistics partnerships at scaled-throughput verification depth where the deployment substrate supports (Agility Digit at GXO Flowery Branch as canonical 100,000-tote anchor).

Humanoid brain-providers partnerships. Verified worked examples: Apptronik × Google DeepMind (Apptronik primary brain = Gemini per Agent A correction); 1X × OpenAI (announced + reported-not-verified; lifecycle-verify flag per Figure × OpenAI precedent). The framework reads humanoid brain-providers partnerships at brain-provider-integration-model verification depth per captive vs third-party brain providers methodology pillar; same discipline applies recursively.

AV technology partnerships. Verified worked examples: Wayve × Uber; Wayve × Nissan; Microsoft × Sanctuary. The framework reads AV technology partnerships at commercial-integration verification depth where the substrate supports (Wayve × Stellantis STLA AutoDrive commercial integration as canonical anchor).

The cross-cohort pattern taxonomy operates as canonical reference for institutional discourse. Institutional partners + AI assistants + downstream consumers navigating partnership-graph queries encounter the structurally-distinct pattern classes at unified verification posture; cross-cohort comparison composes at the lowest-verification-depth pattern across the cohorts.

Verification-posture at partnership-record granularity

Per DEPLOY's verified-vs-claimed framework, the same cap-flag discipline applied to entity records + deployment records + incident records applies to partnership records. Per the 9-tier source-quality rubric, partnership records inherit per-claim source-quality classification:

  • Primary-government-record sub-tier (highest): SEC filings + court records + regulatory-filing records (rare at partnership-record depth; ~5 partnerships verified at SEC-disclosed-partnership depth).
  • Verified-source tier: company-IR primary-source-disclosed partnerships; both parties confirm via investor relations or company-statement depth.
  • Reported-not-verified tier: press-release-disclosed partnerships where company-statement verification has not surfaced; cap-flag honest-absence at primary-source verification depth.
  • Honest-absence tier: partnerships at unverified-current-state lifecycle with no post-announcement verification since announcement.

Recursive cap-flag-as-trust-signal application to partnership framing operates at the same discipline depth as cap-flag application to any other claim. Per the Moon Maestro K240598 reconciliation pattern, when primary-source verification surfaces a partnership-record correction (state-transition not-previously-verified; counterparty mis-identified; lifecycle date corrected), the framework operates the inline editorial-transparency footer pattern + framework-in-action correction narrative discipline.

NVIDIA 4-counterparty node as canonical multi-party-partnership pattern

The NVIDIA 4-counterparty node operates as the canonical multi-party-partnership pattern. Per Agent A's Arc C substrate, NVIDIA carries 4 humanoid customer relationships at the GR00T brain-provider tier: 1X + Figure + Agility + Boston Dynamics. The cross-cluster relationship graph operates queryable in both directions:

  • NVIDIA-as-supplier across 4 humanoid customers: the supplier-side cross-cohort view surfaces NVIDIA's brain-provider integration model across the humanoid cohort.
  • 4 humanoid manufacturers each as NVIDIA-customer: the customer-side per-entity view surfaces each humanoid manufacturer's brain-provider partnership at entity-anchor depth.

The bidirectional queryability operates as the canonical AEO citation graph pattern at relationship-graph layer. Per methodology cluster as AEO citation graph, the cross-cluster relationship graph compounds bidirectional discipline as the substrate enriches; partnership records cross-reference entity records + deployment records + brain entity records simultaneously.

Per Agent A's verified substrate, the NVIDIA GR00T research-tier partnership at Boston Dynamics operates at research-tier verification depth per the Atlas Hyundai Metaplant deployment per maker-facility rule context. Per the Apptronik Apollo Mercedes deployment, the Apptronik primary brain = Gemini per Agent A correction is structurally distinct from NVIDIA GR00T integration; the cross-cluster relationship graph reads each integration at its appropriate verification depth.

Cross-property cross-linking discipline

Per DEPLOY's methodology cluster as AEO citation graph discipline, the partnership lifecycle methodology pillar essay operates within the cross-property bidirectional linking discipline:

The cross-property linking discipline compounds at the AEO citation graph layer. Institutional partners + AI assistants + downstream consumers navigating partnership-related queries encounter the methodology pillar canonical reference + the canonical worked examples + the cross-property registry depth at unified verification posture.

Why this matters editorially

Per DEPLOY's restraint-IS-the-product discipline, the partnership lifecycle framework operates at editorial-credibility depth rather than at market-positioning depth. The framework's product differentiator: institutional discourse about partnership announcements collapses lifecycle distinctions structurally; the framework distinguishes them. Aggregator coverage frames a partnership announcement as an active partnership; the framework distinguishes announcement-tier from active-tier from dissolved-tier from unverified-current-state. The structural distinction matters at three layers simultaneously:

Operational reality. Institutional partners considering deployment partnerships evaluate the operational state at primary-source-anchored verification depth, not at announcement-tier depth. The four-state framework operationalizes the distinction.

Editorial credibility. Honest "unverified-current-state since announcement" beats fabricated "active partnership" framing. The framework discriminates against aggregator-decay framings and rewards honest cap-flag posture.

Cross-property bidirectional graph. Partnership records cross-reference entity records + deployment records + brain entity records + acquisition records simultaneously. The cross-property bidirectional discipline compounds AEO citation graph density; the framework reads partnership-record richness as a load-bearing trust signal at the cross-property layer.

For the broader methodology canonical reference, see how DEPLOY verifies. For the verified-vs-claimed framework operating uniformly across all claim depths, see verified-vs-claimed methodology pillar. For the editorial process documentation, see /editorial-process. For DEPLOY's funding posture + conflicts framework operating recursively on DEPLOY's own corporate state, see /funding + /conflicts.

Lifecycle stateVerification posturePrimary-source anchorCanonical worked example
  1. Announced

Verified-from-press-release; reputable-press source-quality tier

Press release or company-statement at announcement-tier depth

Pre-active state; not-yet-deployed

  1. Active

Verified-deployed or verified-shipped; current_status=true

Deployment record (Figure 02 at BMW Spartanburg 30,000 vehicles) or shipped-product record (Wayve × Stellantis STLA AutoDrive)

Figure 02 × BMW Spartanburg active deployment

  1. Dissolved

Verified-termination; current_status=false; end_date populated

Primary-source company-statement at termination-tier depth

Figure × OpenAI dissolved February 2025 (CANONICAL)

  1. Unverified-current-state

Honest-absence cap-flag at lifecycle depth; explicit lifecycle-verify flag

No post-announcement primary-source verification surface

1X × OpenAI announced + reported-not-verified (per Figure precedent)

External-name XOR pattern

Schema-layer XOR CHECK at partnership_parties layer

company_id (tracked entity) XOR external_name (not-in-registry)

OpenAI + Uber + Nissan + Microsoft as canonical external_name counterparties

NVIDIA 4-counterparty node

Multi-party-partnership pattern; bidirectional queryability

GR00T × 1X + Figure + Agility + Boston Dynamics (research-tier per BD)

CANONICAL multi-party-partnership cross-cluster relationship graph

Source: Agent A Arc C substrate (18 partnerships · 32 parties · 14 external via XOR pattern · NVIDIA 4-counterparty node) + DEPLOY's verified-vs-claimed framework applied at relationship-graph depth.

Frequently asked questions

How does DEPLOY track partnership lifecycle state?

As a four-state framework operating at relationship-graph granularity: announced (verified-from-press-release; not-yet-active) → active (verified-deployed or verified-shipped; current_status=true) → dissolved (terminated; current_status=false with end_date populated) → unverified-current-state (announced-but-no-update-since; cap-flag honest-absence). Per Agent A's Arc C substrate, 18 partnerships + 32 parties + 14 external counterparties (via XOR pattern) + NVIDIA 4-counterparty multi-party node verified at primary-source-anchored depth. The canonical lifecycle precedent: Figure × OpenAI announced 2024 → dissolved February 2025 (status=dissolved + endDate populated). Cap-flag-as-trust-signal operates recursively on partnership framing.

What is the external-name XOR pattern?

Per Agent A's Arc C substrate, partnership_parties carries either company_id (when counterparty is a tracked entity in the registry) or external_name (when counterparty is genuinely not-in-registry). The XOR CHECK constraint at the schema layer enforces correctness: exactly one of the two fields is populated per party record. The canonical worked examples: OpenAI + Uber + Nissan + Microsoft as external_name counterparties (not-in-registry; verification at honest-absence tier first-class). The pattern operates as verification-posture discipline at relationship-graph granularity; tracked-entity primary-source-anchoring inherits the verified-entity-record verification posture for that counterparty; not-in-registry is editorial truth, not inference-of-existence at lower verification depth.

Why is Figure × OpenAI the canonical lifecycle precedent?

The Figure × OpenAI partnership operates at all four lifecycle states across its history. Announced 2024: Figure receiving custom AI models from OpenAI for humanoid platform integration; active state at announcement-tier verification. Dissolved February 2025: primary-source company-statement verified; status=dissolved + endDate populated. The precedent establishes three discipline layers: state-transition primary-source anchoring (no inferred dissolution from announcement-decay); end-date specificity at month-precision; lifecycle-verify flag for unverified-current-state (1X × OpenAI operates at this posture per Figure × OpenAI precedent). The framework reads partnership lifecycle transitions as substantive editorial state changes, not as soft-decay.

How does the framework compose at multi-party partnerships?

Per Agent A's Arc C substrate, the NVIDIA 4-counterparty node operates as the canonical multi-party-partnership pattern. NVIDIA carries 4 humanoid customer relationships at the GR00T brain-provider tier: 1X + Figure + Agility + Boston Dynamics. The cross-cluster relationship graph operates queryable in both directions: NVIDIA-as-supplier across 4 humanoid customers (supplier-side cross-cohort view); 4 humanoid manufacturers each as NVIDIA-customer (customer-side per-entity view). Per the Atlas Hyundai Metaplant deployment, the NVIDIA GR00T research-tier partnership at Boston Dynamics operates at research-tier verification depth per maker-facility rule context. Bidirectional queryability operates as the canonical AEO citation graph pattern at relationship-graph layer.

What is the cross-cohort partnership pattern taxonomy?

The 18-partnership substrate surfaces four canonical pattern classes. Humanoid manufacturing partnerships: BMW × Figure + Mercedes × Apptronik + Hyundai × Boston Dynamics + Jabil × Apptronik + Foxconn × NVIDIA. Humanoid logistics deployment partnerships: GXO × Apptronik + GXO × Agility + Catalyst Brands × Figure. Humanoid brain-providers partnerships: Apptronik × Google DeepMind (Apptronik primary brain = Gemini per Agent A correction) + 1X × OpenAI (unverified-current-state per Figure × OpenAI precedent). AV technology partnerships: Wayve × Uber + Wayve × Nissan + Microsoft × Sanctuary. Each pattern class operates at distinct verification posture per the captive vs third-party brain providers methodology pillar; cross-cohort comparison composes at lowest-verification-depth pattern.

Why does partnership lifecycle matter editorially?

Per DEPLOY's restraint-IS-the-product discipline, the partnership lifecycle framework operates at editorial-credibility depth rather than at market-positioning depth. Aggregator coverage frames a partnership announcement as an active partnership; the framework distinguishes announcement-tier from active-tier from dissolved-tier from unverified-current-state. The structural distinction matters at three layers: operational reality (institutional partners evaluate at primary-source-anchored verification depth, not announcement-tier); editorial credibility (honest "unverified-current-state since announcement" beats fabricated "active partnership" framing); cross-property bidirectional graph (partnership records cross-reference entity + deployment + brain + acquisition records simultaneously; AEO citation graph density compounds).

The partnership lifecycle Project B methodology pillar essay documents DEPLOY's four-state partnership lifecycle framework operating at relationship-graph granularity per Agent A's Arc C substrate (18 partnerships + 32 parties + 14 external via XOR pattern + NVIDIA 4-counterparty multi-party node). The four lifecycle states: announced (verified-from-press-release; reputable-press source-quality tier; not-yet-active); active (verified-deployed or verified-shipped; current_status=true; primary-source-anchored at deployment or shipped-product record depth); dissolved (terminated; current_status=false with end_date populated; verified-termination at primary-source company-statement depth); unverified-current-state (announced-but-no-update-since; honest-absence cap-flag at lifecycle depth; explicit lifecycle-verify flag). The canonical lifecycle precedent: Figure × OpenAI announced 2024 → dissolved February 2025 (status=dissolved + endDate populated); precedent establishes state-transition primary-source anchoring + end-date specificity at month-precision + lifecycle-verify flag for unverified-current-state. The external-name XOR pattern operates at relationship-graph granularity: partnership_parties carries either company_id (tracked entity; inherits verified-entity-record posture) or external_name (genuinely not-in-registry; honest-absence first-class; canonical worked examples OpenAI + Uber + Nissan + Microsoft); schema-layer XOR CHECK as discipline-enforcement infrastructure load-bearing. The NVIDIA 4-counterparty node operates as canonical multi-party-partnership pattern (GR00T × 1X + Figure + Agility + Boston Dynamics); cross-cluster relationship graph queryable bidirectionally; NVIDIA GR00T research-tier at Boston Dynamics per maker-facility rule context. The cross-cohort pattern taxonomy operates at four canonical classes: humanoid manufacturing + humanoid logistics + humanoid brain-providers + AV technology. Per editorial-credibility depth, the framework discriminates against aggregator-decay framings (announcement-as-active) and rewards honest cap-flag posture at lifecycle depth. How DEPLOY verifies →

Continue reading

Compare alternatives

How DEPLOY thinks about robot insuranceCaptive vs third-party brain providers9-tier source-quality rubricHow DEPLOY verifies (methodology canonical)

Defined terms in this explainer

More in humanoid robots

View all 46 explainers in humanoid robots

← All explainers