FleetWorks
One station to many, without centralizing control
Compare station readiness with redacted evidence summaries while every station stays the local authority — no raw inspection data is centralized and no station is promoted automatically.
Step 01
Fleet dashboard
Compare stations using redacted readiness summaries — never raw station data.
Inputs
- Station readiness summaries
- Evidence completeness
- Coverage completeness
- Review state
- Drift state
- Commissioning state
- Ops metrics
Proof generated
- Fleet posture snapshot
- Station summary hashes
- Redaction receipt
- Comparison timestamp
Where it appears in the app
- FleetWorks
- Ops metrics
- Stations
- Capabilities
AI Sense support
- Detects fleet-level risk
- Ranks stations needing a human look
- Explains the readiness spread
Safety boundary
- The fleet dashboard cannot command, upgrade, or promote any station.
Step 02
Readiness by station
Each station keeps its own readiness state and its own blocker trail.
Inputs
- Station registry
- Commissioning status
- Review status
- Evidence completeness
- Known limitations
- Signed-pack state
Proof generated
- Station readiness rollup
- Blocker list
- Readiness reason codes
Where it appears in the app
- Stations
- Commissioning
- SignedPacks
- FleetWorks
AI Sense support
- Explains why one station is ready for review while another stays blocked
- Flags a missing readiness reason
- Flags stale registry data
Safety boundary
- Fleet readiness does not promote, unlock, or activate a station.
Step 03
KPI comparison
Compare station KPIs side by side without mixing local truth or centralizing evidence.
Inputs
- Evidence metrics
- Coverage metrics
- Review metrics
- False-reject candidates
- Escape candidates
- Station health
Proof generated
- KPI comparison receipt
- Per-station source references
- Redacted metric summary
Where it appears in the app
- Review analytics
- EvidenceWorks
- FleetWorks
AI Sense support
- Finds outlier stations
- Explains likely missing proof or process gaps
- Ranks which station needs attention first
Safety boundary
- KPI comparison cannot change verdicts, QA decisions, or station labels.
Step 04
Drift comparison
Compare drift posture across stations and prioritize recalibration attention for humans.
Inputs
- Vision Twin drift summaries
- Calibration status
- Lighting measurement status
- Coverage stability
- Station health
Proof generated
- Fleet drift rollup
- Attention priority
- Drift comparison receipt
Where it appears in the app
- Vision Twin fleet
- Vision Twin knowledge
- Vision Twin work package
- FleetWorks
AI Sense support
- Ranks drift risk across stations
- Prepares a manual calibration work package
- Explains the recommended human check
Safety boundary
- Drift comparison cannot recalibrate, restore claims, or activate packs.
Step 05
Risk register
Fleet expansion is blocked by unresolved station risk — no risk is accepted automatically.
Inputs
- Escape candidates
- Critical risks
- Known-bad miss events
- Coverage blockers
- Commissioning exceptions
- Governance blockers
Proof generated
- Fleet risk register
- Risk severity
- Risk owner reference
- Evidence references
Where it appears in the app
- Governance
- Commissioning
- EvidenceWorks
- FleetWorks
AI Sense support
- Groups repeated risks
- Identifies missing evidence before expansion
- Explains each risk severity
Safety boundary
- The risk register cannot accept risk automatically.
Step 06
Rollout plan
Rollout plans are administrative guidance for humans — not station commands.
Inputs
- Readiness by station
- Signed-pack status
- Commissioning state
- Risk register
- Pilot results
- Customer constraints
Proof generated
- Rollout plan draft
- Station-specific blockers
- Required signoffs
- Rollback requirement
Where it appears in the app
- Improvements
- SignedPacks
- Commissioning
- FleetWorks
AI Sense support
- Suggests a rollout order
- Explains why a station should stay blocked or shadowed
- Flags a missing signoff
Safety boundary
- A rollout plan cannot skip commissioning, signoffs, station locks, or validation.
Step 07
Executive summary
Summaries help humans decide — they do not approve anything for production.
Inputs
- Fleet posture
- Station readiness
- Risk register
- Drift summary
- Go / no-go state
- Known limitations
Proof generated
- Executive summary
- Omissions list
- Evidence reference list
- Summary hash
Where it appears in the app
- SignalOps
- FleetWorks
- Trust
- Customer packets
AI Sense support
- Removes unsupported claims
- Highlights missing proof
- Prepares a customer-safe summary
Safety boundary
- The executive summary cannot approve production or change any station.
Step 08
Go / no-go by station
Each station receives its own recommendation and its own blocker trail.
Inputs
- Evidence completeness
- Coverage completeness
- Review state
- Drift state
- Known limitations
- Customer constraints
Proof generated
- Go / no-go station recommendation
- Reason trail
- Blocker list
- Known-limitation summary
Where it appears in the app
- EvidenceWorks
- Commissioning
- FleetWorks
- SignalOps
AI Sense support
- Explains why each station has a different recommendation
- Flags a missing reason code
- Highlights an open blocker
Safety boundary
- A go / no-go recommendation is not approval and not machine authority.
Step 09
Known limitations
FleetWorks includes what it cannot know or control — raw evidence stays local.
Inputs
- Redaction state
- Station authority state
- Raw-data boundary
- Known missing proof
- Model limitations
- Sampling limitations
Proof generated
- Limitations ledger
- Redaction receipt
- Local-authority statement
Where it appears in the app
- Trust
- Governance
- FleetWorks
AI Sense support
- Prevents confidence inflation
- Flags an unsupported fleet-level claim
- States where the fleet view may lag station truth
Safety boundary
- FleetWorks receives redacted summaries only; raw station evidence never leaves the station.
AI Sense finds the next human check, never approves
AI Sense
One reading layer across every FleetWorks step
Observes evidence, finds missing proof, explains uncertainty, ranks human checks, and prepares handoffs — it never commands hardware.
Reads
- Evidence bundles
- Review events
- QA decisions
- Vision Twin drift
- Commissioning blockers
- Governance decisions
- Station registry
- Ops metrics
Produces
- Findings
- Evidence-gap warnings
- Work-package hints
- Commissioning questions
- Support summaries
Never
- No PLC writes
- No force PASS
- No recovery clear
- No robot commands
- No camera/light commands
- No production approval
- No evidence mutation
- No QA decision mutation
AI Sense observes evidence and guides humans — it records nothing and changes nothing. It does not command a station, write a PLC, clear recovery, reset safety, force a pass, approve production, sign off, or mutate any review, QA decision, commissioning, governance, evidence, or runtime state. Every recommendation is a suggestion for a human to carry out; the PLC and safety circuit remain authoritative.
Customer-safe executive summary
Proof by reference, never raw data
The redacted executive summary carries fleet posture, per-station status, ranked risks, the drift summary, a go/no-go recommendation per station, and an explicit omissions list — so a customer can review the fleet without any station ever handing over raw internals.
- fleet_posture
- calm posture across the compared stations
- station_status
- per-station readiness by reference, local labels only
- top_risks
- ranked risk-register entries (severity + owner role)
- drift_summary
- attention priority + recommended human check
- go_no_go_by_station
- per-station recommendation + reason trail
- known_limitations
- what the fleet view cannot know or control
- omissions
- explicit list of what was withheld
- summary_hash
- integrity fingerprint of the redacted summary
The summary never contains raw images or evidence frames, raw PLC coils or registers, private keys or signing secrets, authority tokens, operator personal identity, local file paths or command payloads.
Signed-in teams run this operationally in the HoldField app, under FleetWorks — where fleet posture, readiness by station, KPI and drift comparison, the risk register, a rollout draft, per-station go/no-go, and the redacted executive summary are compared from redacted summaries only, and where every station stays the local authority: nothing here promotes, unlocks, activates, or commands a station. Open the workspace →