One governed path: physical signal is perceived, interpreted, resolved against operating context, decided under policy, then recorded as an authorized action with lineage held from sensor frame to signature.
Operational + physical AI · Mission-critical systems
From signal to accountable action. Operational and physical AI that turns enterprise data and real-world signals into governed decisions.
Request a briefingApplications
Start with the operating question. Connect the evidence. Put a useful decision in front of the person accountable for it.
Delivery, cost and risk are moving apart on one program.
Schedule, financials and staffing all contribute. Contracts and risk do not.Review the variance with its schedule, cost and staffing context still attached.
Route the follow-up to the accountable owner.One review path replaces disconnected schedule, cost, contract, and risk checks.
Program Performance Cockpit
Connect schedules, financials, contracts, staffing, and risk so leaders can review variance against the operating baseline before the next program decision.
Business / executive
Connect schedules, financials, contracts, staffing, and risk so leaders can review variance against the operating baseline before the next program decision.
One review path replaces disconnected schedule, cost, contract, and risk checks.
Review the decision with the source context still attached.
Explore Program Performance Cockpit
Industrial / infrastructure
Relate work history, condition signals, inspections, and parts to asset criticality so maintenance teams can challenge and prioritize the next action.
Maintenance priority reflects condition, consequence, work history, and readiness together.
Move a traceable recommendation into the maintenance plan.
Explore Maintenance Intelligence
Field / readiness / logistics
Give field teams the job plan, asset history, procedures, and current exceptions at the point of work, then route findings back to the accountable owner.
The point of work gains current context while findings return to the accountable owner.
Keep the field decision and its closeout connected.
Explore Field Operations Copilot
Facilities / portfolio
Bring occupancy, maintenance, finance, and compliance into one portfolio view so operators can trace variance to sites, assets, and operating conditions.
Portfolio variance becomes a traceable site-level operating question.
Focus the next intervention on the driver, not the symptom.
Explore Portfolio Intelligence
Illustrative
FLAGSHIP APPLICATION
The operating system for Operational + Physical AI across eight industries.
Connect enterprise evidence, physical signals and authorized action.
Start with the diagnosticESTABLISHEDSelected into Palantir Startup Fellowship Cohort 003.Read the announcement
Industries
Each industry brings its own decisions, constraints and physical reality.
Explore all industries →What should the team inspect or maintain next?
Explore Industrial OperationsWhat should the team inspect or maintain next?
Explore industry ↗Defense & ReadinessWhich gap threatens the mission, and who can close it?
Explore industry ↗Critical InfrastructureWhere does operational consequence change the priority?
Explore industry ↗Energy & UtilitiesHow should crews, assets, and conditions reshape the response?
Explore industry ↗Facilities & Field OperationsWhat does the person at the point of work need now?
Explore industry ↗Logistics & Supply ChainWhere are supply risk and operating urgency out of alignment?
Explore industry ↗Commercial OperationsWhat changed, and what does it do to the plan?
Explore industry ↗Real Estate & Property ManagementWhich sites are drifting, and what is driving it?
Explore industry ↗Why Ridgeline
Enterprise evidence. Physical context. Human authority.
Approved cameras, lidar, thermal, acoustic, vibration, GNSS, and OT/ICS telemetry can be captured where the work happens.
Edge inference can turn raw signal into detections, states, and anomalies before anything leaves the site, where the deployment requires it.
Detections can resolve against the mission ontology - the asset, the crew, the location, the procedure, the risk - when source mappings are approved.
Reasoning is bounded by policy and evidence, and produces a recommendation an operator can interrogate.
Approved actions can route into systems of record with lineage from sensor frame to signature, when writeback is in scope.
Physical signal→Governed evidence→Authorized action
Environment, assets, and OT signal → Time-stamped raw capture
Fixed and mobile sensing · Mobile capture · OT / ICS signal · Inspection media
Raw capture → Detections, states, anomalies
Computer vision · Anomaly detection · Sensor fusion · On-device inference
Detections → Resolved asset and mission context
Asset resolution · Spatial context · Condition history · Procedure linkage
Context and policy → Evidence-backed recommendation
Bounded autonomy · Evidence-backed · Operator review · Escalation paths
Authorized recommendation → Record of action with lineage
Work order write-back · Task dispatch · Audit trail · Replay and reversal
Built for the operating decision, not generic chat.
Connect records and signals to the people doing the work.
Field conditions shape decisions about assets and crews.
Source context, permissions and human review stay connected.
Conceptual operating model. Sector coverage is not a claim of deployed customer results.
Sovereignty · who owns what
Keep your data, models and decision rights portable across seven governed layers.
Review the sovereignty model →Operational records → Data location, retention and access remain explicit
Seven layers build from data boundary through compute, model choice, semantic definitions, workflows and decision rights to a retained operating record.
Operational records → Data location, retention and access remain explicit
Ownership · Data location · Export
Approved environments → Workloads execute inside the chosen boundary
Hosted · Dedicated · Your cloud
Client-selected models → Models remain replaceable inputs, not the record
Portable · Versioned · Replaceable
Metrics, entities and rules → Meaning is governed outside any single model
Ontology · Definitions · History
Context and operating procedure → Recommendations follow bounded, reviewable paths
Context · Procedure · Lineage
Roles, permissions and thresholds → Consequential action stays with authorized people
Authority · Approval · Audit
Sources, decisions and outcomes → A portable record the organization retains
Evidence · Decisions · Outcomes
Start with one operating decision
Tell us what must change, who owns the outcome, and where the evidence lives. We’ll start there.
