John Deere + DirtFleet.

John Deere Operations Center is the canonical surface for machine performance, location history, and utilization analytics — built by JD, owned by JD, deeply integrated with the OEM equipment. DirtFleet handles what Operations Center doesn't: maintenance scheduling, hour-based PM triggers, work-order queues for the mechanic shop, tool tracking, repair-log P&L by project. The two systems live in different parts of the workflow; the integration glues them together so you don't re-key data between them.

1. Pattern 1 — Operations Center OAuth + machine sync (live)

   The DirtFleet OAuth client for John Deere Operations Center is shipped (lib/integrations/john-deere). Authorize once, DirtFleet pulls your machine roster + engine-hours on a 15-minute polling cadence. Each new machine becomes an Asset; each hours reading writes a HoursLog.

   When to pick this

   If your fleet is heavy on JD iron (combines, sprayers, tractors, construction equipment) and you already use Operations Center for telematics + utilization tracking. The maintenance workflows live in DirtFleet; OC stays the OEM data source.

   Show the setup

   # Step-by-step from a customer perspective — no code on your end:
   #
   # 1. In DirtFleet:  /settings/integrations/john-deere-operations-center
   # 2. Click "Connect Operations Center" — redirects to JD's OAuth page
   # 3. Approve the requested scopes:
   #      - Read Machine Inventory  (so we can mirror your roster)
   #      - Read Machine Hours      (so we can sync engine hours)
   #      - (Optional) Read Service Codes  (for DTC routing — Pattern 3)
   # 4. JD redirects back; DirtFleet stores the refresh token encrypted
   # 5. Initial sync runs immediately — typically completes in 1-3 minutes
   #    for a 50-machine fleet
   # 6. Subsequent polls run every 15 minutes; each one is a delta-only
   #    upsert so the API call cost stays flat

2. Pattern 2 — AEMP-generic fallback (covers OEM brands beyond JD)

   DirtFleet also speaks the AEMP (Association of Equipment Management Professionals) telematics XML standard, which is the de-facto cross-vendor protocol for off-road equipment. If you have JD + Cat + Komatsu in the same fleet, the AEMP feed normalizes them all.

   When to pick this

   Mixed-OEM fleets that don't want one integration per brand. AEMP gives you a single endpoint per OEM that all return the same XML shape; DirtFleet's AEMP adapter parses it identically regardless of source.

   Show the setup

   # Each OEM has its own AEMP endpoint; the data shape is the same.
   # DirtFleet stores the credentials at /settings/integrations/aemp and
   # polls each registered endpoint on a 30-minute cadence.
   #
   # What we receive (from any AEMP source):
   #   <Fleet>
   #     <Equipment>
   #       <SerialNumber>...</SerialNumber>
   #       <Make>...</Make>
   #       <Model>...</Model>
   #       <CumulativeHours>...</CumulativeHours>
   #       <Location><Latitude>...</Latitude>...</Location>
   #       <Odometer>...</Odometer>
   #     </Equipment>
   #   </Fleet>
   #
   # What we do with it:
   #   - SerialNumber → Asset.serial (matched 1:1)
   #   - Make+Model   → equipmentSnapshot JSON
   #   - CumulativeHours → HoursLog (deltas only)
   #   - Location     → optional yard reassignment if a yard sits within 200m
   #
   # AEMP is documented at https://aemp.org/standards/.
   # OEMs that publish: JD, Cat, Komatsu, Hitachi, Volvo CE, Doosan,
   # Bobcat, Bell, Bell-Hyundai, etc.

3. Pattern 3 — JD DTC routing (roadmap)

   Operations Center surfaces Service Advisor codes from the machine's CAN bus. DirtFleet's DTC-narrative AI translates the code into plain English and creates a YELLOW flag with context already attached — same shape as the Samsara DTC routing.

   When to pick this

   Operations that want pre-triaged shop alerts. Mechanic gets the code + a paragraph explanation + likely cause + suggested next step before they walk to the machine.

   Show the setup

   # Status: Roadmap — gated on activating customer.
   #
   # The AI narrative pipeline is shipped (lib/ai-dtc-narrative.ts;
   # used today by drivers logging DTCs manually). The OC webhook
   # subscription + automated flag creation is the wiring not yet
   # in place; the OC OAuth scope is the dependency (re-prompt the
   # customer if they didn't grant Read Service Codes initially).
   #
   # Once active, the flow is:
   #   1. JD Operations Center sends a Service Advisor event webhook
   #   2. DirtFleet authenticates the event signature
   #   3. AI narrative generates the human-readable context
   #   4. We post a YELLOW flag with the narrative + machine reference
   #   5. Mechanic dashboard sorts it by urgency and surfaces in queue
   #
   # Email hello@dirtfleet.app to be the first OEM-routing customer.

• Bidirectional sync. Operations Center → DirtFleet is one-way. Maintenance state lives in DirtFleet; the OC view stays read-only from our side.
• Operator/yield analytics. Stays in Operations Center. That's the precision-ag domain we're not in.
• Real-time location streaming. 15-minute polling is sufficient for maintenance triggers; sub-minute location is OC's product.

See also /integrations/samsara (on-road) and /integrations/geotab (mixed). For mixed-OEM off-road fleets (JD + Cat + Komatsu), the AEMP fallback (Pattern 2 here) is the right single-integration path.