How did DEPLOY correct the Figure 03 BMW deployment narrative?

The framing error

Aggregator coverage of Figure AI's humanoid deployment narrative frequently frames it as "Figure 03 deployed at BMW Plant Spartanburg + Leipzig." Two BMW plants. Two deployments. Both attributed to the current-generation Figure 03 humanoid platform.

The framing is wrong in three distinct ways at once.

Per primary-source verification, the actual deployment state involves three distinct entities and three distinct deployments:

• BMW Spartanburg deployment: Figure 02, NOT Figure 03. The previous-generation Figure 02 humanoid completed a verified 11-month deployment from August 2024 producing 30,000 BMW X3 vehicles + 1,250 hours of runtime handling more than 90,000 parts in chassis-assembly tasks. The Spartanburg deployment is the canonical commercial humanoid manufacturing reference for the cohort with end-product OEM acceptance verification. It happened. It was Figure 02. Not Figure 03.
• BMW Leipzig deployment: Hexagon AEON, NOT Figure. BMW Group's Plant Leipzig humanoid production expansion went to Hexagon AEON per Agent A primary-source verification. The aggregator framing attributing Leipzig to Figure operates outside primary-source-anchored verification of the BMW Leipzig deployment-partner identity.
• Figure 03 current deployment: Catalyst Brands Reno Distribution Logistics Center, NOT BMW. The current-generation Figure 03 has one confirmed customer deployment as of mid-2026: the Catalyst Brands Reno distribution logistics center, announced through Figure's May 2026 partnership signal. The Figure 03 commercial-pilot positioning operates at Catalyst Brands logistics customer scope, not at BMW automotive assembly scope.

The aggregator framing collapses three distinct facts into one wrong claim about a fourth entity. Per DEPLOY's framework discipline, audit-first verification + aggregator-drift rejection produced the corrected attribution.

---

The audit-first verification pattern

Per DEPLOY's standing operational discipline, deployment-attribution claims operate at primary-source-verified depth. The audit-first verification pattern for humanoid deployment claims:

• Who is the maker (humanoid generation + manufacturer)? Verify maker + product-generation against primary-source-anchored references (manufacturer announcements + product-specific deployment documentation).
• Who is the customer (deployment site + operational scope)? Verify customer + site against primary-source-anchored references (customer announcements + site-specific deployment confirmation + verified operational data).
• What is the deployment scope (verified scale + operational data)? Verify scope at primary-source depth (production-count + runtime + parts-handled + multi-month operational records).
• What is the timing (deployment window + announcement dates)? Verify timing at primary-source depth (announcement date + deployment start + ongoing-vs-completed state).

Running the audit pattern on Figure deployment claims surfaces the maker + customer + scope + timing distinctions explicitly. Each fact has its own primary-source anchor; conflations between facts get caught structurally.

Per the Figure 02 BMW Spartanburg deployment deep-dive, the Spartanburg deployment was Figure 02 (previous generation; bipedal humanoid with proprietary AI stack); BMW Group customer + Spartanburg, South Carolina plant site; 30,000 BMW X3 vehicles + 1,250 hours runtime + 90,000 parts handled scope; 11-month deployment from August 2024 timing. All four audit-pattern axes resolve to Figure 02 at Spartanburg with verified scope + timing.

Per the Figure 03 entity anchor, the Figure 03 generation is current-generation; Catalyst Brands customer + Reno Distribution Logistics Center site; pilot-stage scope (throughput data + multi-customer expansion not yet at Figure 02 Spartanburg disclosure depth); May 2026 partnership announcement timing. The audit pattern resolves Figure 03 to Catalyst Brands Reno, not BMW.

---

The aggregator-drift pattern: three-way conflation

The Figure 03 BMW framing operates as a worked example of a structurally distinct aggregator-drift pattern: cross-customer + cross-generation + cross-maker conflation in a single wrong claim.

The aggregator framing typically reads: "Figure 03 deployed at BMW Plant Spartanburg + Leipzig." The framing implicitly asserts:

1. Figure 03 is at BMW Spartanburg (wrong; was Figure 02 generation)
2. Figure 03 is at BMW Leipzig (wrong; went to Hexagon AEON)
3. Both BMW plants run the same humanoid platform (wrong; one was Figure 02, the other is Hexagon AEON, and Figure 03 is at neither)

Three wrong facts in one sentence. Each fact wrong at a different layer:

• The Spartanburg fact wrong at the generation-distinction layer (collapsing Figure 02 into Figure 03).
• The Leipzig fact wrong at the maker-distinction layer (collapsing Hexagon AEON into Figure).
• The dual-plant fact wrong at the cross-customer-conflation layer (treating two distinct customer-and-site deployments as one).

Per cap-flag-as-trust-signal, the aggregator-drift rejection operates at all three layers simultaneously. The framework reads each fact at its own primary-source verification depth; collapsing across distinctions gets caught structurally.

---

The verified state

The corrected attribution per primary-source verification:

Figure 02 at BMW Spartanburg (verified, completed 11-month deployment per Figure 02 BMW Spartanburg deployment deep-dive):

• Figure 02 humanoid platform (previous generation; not Figure 03)
• BMW Group customer at Spartanburg, South Carolina plant site
• 30,000 BMW X3 vehicles produced + 1,250 hours of runtime + 90,000 parts handled in chassis-assembly tasks
• 11-month deployment window from August 2024 with end-product OEM acceptance verification

Hexagon AEON at BMW Leipzig (verified per Agent A; not Figure):

• Hexagon AEON humanoid platform (Hexagon manufacturer; not Figure)
• BMW Group customer at Plant Leipzig site
• Specific deployment scope + timing operate at lower verification posture pending Hexagon + BMW primary-source confirmation
• The fact that matters: not Figure. The aggregator framing collapsing this into "Figure at BMW Leipzig" is wrong at the maker-distinction layer.

Figure 03 at Catalyst Brands Reno (verified per Figure's May 2026 partnership announcement):

• Figure 03 humanoid platform (current generation; not Figure 02)
• Catalyst Brands customer at Reno Distribution Logistics Center site
• Pilot-stage deployment scope; specific throughput data + multi-customer expansion not yet at Figure 02 Spartanburg disclosure depth
• May 2026 partnership announcement timing per Figure-Catalyst Brands partnership signal

Three distinct deployments. Three distinct customer-and-site pairings. Two distinct humanoid platforms (Figure 02 + Hexagon AEON). The verified state surfaces all three honestly with primary-source verification depth at each.

---

Why the catch matters

The Figure 03 BMW correction operates as a worked example of three-way aggregator-drift rejection at editorial-anchor depth. The catch demonstrates the framework discipline operating at multiple verification layers simultaneously:

• Generation-distinction discipline: Figure 02 vs Figure 03 are distinct humanoid generations with distinct deployment customers + scope + timing. Aggregator framing collapsing the two operates outside primary-source-anchored verification of the generation-customer-mapping.
• Maker-distinction discipline: Figure vs Hexagon AEON are distinct humanoid manufacturers. Aggregator framing attributing BMW Leipzig to Figure operates outside primary-source-anchored verification of the BMW Leipzig deployment-partner identity.
• Cross-customer-conflation rejection: distinct customer-and-site deployments (BMW Spartanburg + BMW Leipzig + Catalyst Brands Reno) get verified independently; conflating across customer-and-site distinctions gets caught structurally.

Per DEPLOY's how-deploy-verifies methodology editorial, aggregator-drift rejection at the deployment-attribution layer is canonical worked example of framework discipline. Trade-press coverage of humanoid deployment narratives compounds errors structurally: a generation conflation in one piece + a maker conflation in another piece + a customer-and-site conflation in a third piece together produce the "Figure 03 at BMW Spartanburg + Leipzig" framing that surfaces across aggregator-driven coverage. The framework discipline catches each layer at primary-source-verification depth; the corrected attribution surfaces honestly.

---

The Figure 03 entity anchor surfaces this distinction

The corrected attribution per the Figure 03 entity anchor:

"Figure 03 is the third-generation humanoid robot from Figure AI, a US humanoid manufacturer based in Sunnyvale, California. The robot stands 173 cm (about 5'8"), weighs 61 kg, and is currently in active pilot deployment at a Catalyst Brands distribution logistics center."

"[Figure 03] is the successor to Figure 02 (the generation that produced BMW Spartanburg's 11-month chassis-assembly deployment over 30,000 BMW X3 vehicles)."

"As of mid-2026, Figure 03 has one confirmed customer deployment: the Catalyst Brands Reno Distribution Logistics Center, announced through Figure's May 2026 partnership signal. The deployment is in pilot stage; throughput data, multi-customer expansion, and contractual scope have not been published at the same depth as Figure 02's BMW Spartanburg deployment (which produced verified end-product OEM acceptance data over an 11-month period)."

The entity anchor surfaces the verified Catalyst Brands Reno deployment + the Figure 02 BMW Spartanburg attribution distinction + the pilot-stage Figure 03 scope all in editorial framing that resists aggregator conflation.

The catch demonstrates the discipline operationally. Not because the catch is exceptional. Because the catch is what the discipline does. Aggregator-drift rejection at three layers simultaneously; primary-source verification at maker + customer + scope + timing per the audit pattern; corrected attribution surfacing honestly in the entity anchor + the cluster framework + the methodology editorial canonical reference.

For the canonical Figure 03 entity anchor, see what is Figure 03. For the canonical Figure 02 BMW Spartanburg deployment deep-dive, see Figure 02 at BMW Spartanburg deployment. For the parallel framework-in-action correction-narratives, see 1X Redwood is a captive brain, not humanoid hardware + Monogram Doug Unis CMO/founder, not CEO. For the methodology editorial canonical reference, see how DEPLOY verifies. For the humanoid cluster framework, see the humanoid robots cluster. For methodology canonical references applicable to three-layer aggregator-drift rejection: verified-vs-claimed at within-entity granularity (Figure 02 vs Figure 03 generation distinction at within-Figure granularity) + the 9-tier source-quality rubric (aggregator-drift vs primary-source-anchored verification source classification).