Nestack Agent Care
Industries / Administration / Utilisation agent

Facilities AI agent · Space utilisation

Space Utilisation AI Agent

Attribute each occupancy figure to the method that produced it, mark every zone as measured or unmeasured before a total is formed, and hold the study for the workplace director.

4–6 weeksTypical delivery
Your stackDeployment
Coverage firstZone-level only
Agent CareAfter launch

What this agent does

Measures the estate, never decides about it

In
01

A zone is measured, and the method that measured it is written beside the figure it produced.

02

A sensor drops out, and the zone it served is marked unmeasured rather than counted as empty.

Reason
03

A wing carries no sensors at all, and it stays outside the denominator until somebody walks it.

04

A badge line and a desk sensor disagree, and both readings are kept with the methods behind them.

05

A room panel shows a booking, and whether anybody sat in that room is a separate observation.

Decide
06

A network association is read as presence on a floor, and never as presence at a given desk.

07

A study opens, and its window and its season are fixed before one figure may leave it.

Out
08

A request arrives for attendance by name, and the agent declines and answers at the zone.

09

Execute write actions only inside the approval boundaries agreed during implementation.

Product statement

The agent measures the zones and states its coverage. The workplace director decides what happens to a floor, a lease or an allocation.

Example workflow

One zone, reading to publication

AgentHuman
1Zone readings receivedBadge events, desk sensors, network associations, room panels or audit sheets
2Zone context assembledThe zone, the method that measured it, the period it covers and the coverage it may claim
3Coverage statement draftedThe zone, its readings, its unmeasured areas and completeness
4Controls appliedCoverage checks, method-consistency checks, period checks and completeness confidence
No human action required

Stages 1 to 4 run unaided, and nothing is published at any of them — the agent is measuring, and the workplace lane opens at the completeness gate.

5DecisionSplits at the completeness gate
Coverage sufficient

Goes to the workplace director to publish.

Anything thin

Adds a workplace analyst read first.

Workplace review

The study is held with its coverage, its methods and the zones they were taken from.

Publish · Append readings · Send to workplace review
Published — by the workplace director
6Portfolio and zone records updatedOnly where write access and records policy allow it
7Outcome evaluatedCoverage accuracy, method fidelity, reviewer corrections and what the read found
Corrections

Each workplace correction is counted in the evaluation.

What should not run autonomously

Human approval stays in control

Outside the boundary — human approval required8 items
Recommending that a floor be released.
Deciding which team a zone belongs to.
Publishing a figure without its period.
Reporting attendance at the person.
Automation boundaryAgent acts unaided
Mark every zone as measured or unmeasured before any.
Carry the method that produced a reading alongside the reading.
Fix the window and the season a study covers before it is opened.
Show which zones carried no sensor and were therefore.
Nothing is published or decided except by a named person, inside the agreed boundaries.
Judging whether a wing is genuinely surplus.
Telling the board the estate is under-used.
Setting the target a zone is measured against.
Changes to leases, allocations or headcount.

Example output

One zone, annotated

Our meeting room agent releases the booking nobody used; this record is what one zone measured across one study window.

Study record · single zoneIllustrative example
Zone
Recorded as
Method
Evidence of record
Confidence
Held for
North wing, level three
Measured in part, coverage stated
Desk sensor
Study window, autumn 2026
Held unpublished
The workplace director, by name
As receivedTaken from the desk sensor feed and the audit sheet — it reaches as far as those sources do.
What the record holds Desk sensor feed Manual audit sheet Unmeasured zone list
Why no decision hereReleasing a floor is an estate judgement made by a person, not a model output.
ActionPublishAppend readingsSend to workplace review
What the score decidesA study under the configured threshold picks up an analyst read before the director sees it.

Value

Where AI adds value

The same four claims, placed at the point in the workflow where each one applies.

Where the value landsValue 01 – 04
Every zoneFrom the floor that reports it
03Evidence

Where the evidence is used

A utilisation study is not a workplace strategy: this page measures what the estate carried and states its coverage, and it claims nothing about why a zone was used as it was.

01Approved path

A desk is not a person

Coverage is settled before utilisation is reported, so a zone with no sensor on it goes out marked unmeasured and never as empty.

02Human review

What was checked, and not found

No measurement standard was purchased or read, no vendor sensor-accuracy claim was accepted, no benchmark from an industry survey was carried onto this page, and what the agent may hold about an individual is governed by your own configured policy on personal data.

04Build an evidence trail

The zone, the count observed in it and the manager who read it stay together.

Integrations

Typical integrations

Five system groups connect to the same agent. Which of them are in scope is decided in discovery.

Access and badge sourcesBadge readers · turnstiles
Entry and exit events
Sensors and building systemsDesk sensors · counters
Zone occupancy readings
Network and room panelsWi-Fi controller · panels
Association and room logs

Agent

Space utilisation measurement

Reads the zones
States the coverage
Holds for the director

Portfolio and audit systemsIWMS · audit sheets
Floor plans and zone maps
Observability & evaluationOpenTelemetry · Langfuse
Supported monitoring/evaluation sources

Integration availability depends on the client's existing systems and API access.

Agent controls

Six grates between the model and the director

Six grates laid over each other, the last most closely woven. What drops through is in the map below.

L6 · Outermost — last line of defenceInward → L1 · closest to the model
L6Rollback / safe modeNarrow the agent to measurement and coverage when evaluation or production signals degrade.Roll back
L5Version monitoringTrack model, prompt and coverage rules; a redrawn zone map or a moved sensor changes what an old figure meant.Track
L4TraceabilityRecord each zone, the method behind it, the period it covers and every read of the study.Record
L3Director releaseHold the study for the workplace director; that hold governs publication, not what the estate ought to do next.Gate
L2Method guardrailsTest each zone against the method configured for it; badge, sensor, network and panel readings are never pooled into one number.Restrict
L1Coverage thresholdsRoute a study whose measured share falls short to an analyst read before the director sees it.Require review
Model coreMeasurement assembled — the zone, the method, the period and its coverage
L1 – L2Test whether a study may stand
L3Puts the publication in a person's hands
L4 – L5Keep the zone and the count behind it
L6Narrows to aggregate reporting when signals degrade

How Nestack evaluates it

Evaluate the whole assembly — not only the utilisation figure that comes out.

Coverage runs the whole depth of the workflow, and every layer is cut by slice.

Surface — the study a board reads
Depth of coverage ▼
E1Final-output evaluationDid the figure carry the coverage it was built on?
E2Step-level evaluationDid the agent read the right zone, the right method and the live floor plan?
E3Tool evaluationDid it read and write the correct zone record and the correct study?
E4Confidence calibrationDo thinly covered zones actually attract more workplace corrections?
E5Slice evaluationHow does performance change across specific building types?
E6Business outcomeHow many studies needed a correction before the director published?
Floor — the estate the board answers for

Failure modes

Where each failure originates in the agent

Seven failure modes, each held at the stage where it becomes visible.

Agent lifecycleDirection of processing →
01 · Retrieval1 mode
KN-03

Stale floor plan read

The zone map read is not the one now in use.

Stage gathersThe zones, the methods, the dates and the maps
02 · Reasoning2 modes
KN-04

Utilisation asserted, not covered

A total is formed over a zone nobody measured.

KN-06

Methods mixed across zones

Badge and sensor readings are pooled as one.

Stage proposesThe zones, their methods and coverage
03 · Tool / write2 modes
KN-02

Thin study passed forward

A study moves on without the analyst read.

KN-05

Bound to the wrong zone

A reading is filed against the wrong floor.

Stage writesOnly where write access and approval policy allow it
04 · Output1 mode
KN-01

Published, coverage unstated

A figure lands on the record without the areas it never covered.

Stage returnsThe study a director publishes and a board reads
05 · Change / Version1 mode
KN-07

Silent coverage regression

A sensor estate changes, and the denominator does not.

Stage tracksModel, prompt, coverage rules and study fields
Sev-1 · an unmeasured wing counted empty Sev-2 · methods pooled into one figure Sev-3 · source degrades, study held back

Affected slices

Legacy wings absorb the corrections

A zone-level coverage figure can read clean while sensor-sparse legacy wings carry most of the rework. Nestack reports the correction rate by zone type, not only in total.

Slice performance — reported separately, not only in aggregateIllustrative example
SliceFailure rateLift Lift vs. thresholdStatus
Sensor-sparse legacy wings8.4%3.6× Review
Mixed-method floors6.0%2.6× Review
Meeting-room clusters3.7%1.6× Watch
Fully instrumented floors1.9%0.8× Normal
Bar: correction-rate lift vs. fully instrumented floor baseline · scale 0–4.0× · tick marks the 2.0× review threshold 2 of 4 slices over threshold

Evidence-linked improvement

What an unmeasured wing costs

The loop shuts when the sensor blind spot is a regression case. That suite is what the next study published is measured against.

Improvement cycle · five stagesSwitchback — the path turns at Improve and returns at Learn
01Detect

Correction rate rises on sensor-sparse legacy wings.

02Diagnose

The wing with no sensors that read as empty in every board pack is worked back until one cause remains.

03Improve

Any change goes out numbered, with the zones that caused it attached.

04Verify

One zone case still failing is enough to hold the release back.

05Learn

One case joins the suite, one line joins the measurement record.

Learn → DetectThe return edge. The next study is measured against a suite one case longer.

Typical build scope

Twelve workstreams across six weeks

The build scope read against the delivery timeline. Week structure follows the six-week plan — discovery, sources, measurement assembly, evaluation, integration, then production validation and handover.

Workstream Week 1Week 2Week 3Week 4Week 5Week 6
01Coverage-baseline and automation-boundary work.
02Badge, sensor and floor-plan sources.
03Zone-to-method and coverage-completeness mapping.
04Zone and sensor reading ingestion.
05Zone, method and period binding.
06Coverage scoring and review routing.
07Director publication workflow.
08Workplace-system integration.
09Coverage and aggregation cases.
10Guardrails and publication controls.
11Zone-trail instrumentation.
12Deployment, documentation and Agent Care handover.
12 workstreams · 6 weeks · bar shows the weeks a workstream is active — several run in parallel Final scope and sequence confirmed in discovery

Engagement tiers

What each tier includes

Rows are the capabilities named in each tier's scope. Higher tiers include everything below them.

Capability✓ in scope · — not at this tier PilotOne building, one study ProductionProduction measurement workflow AdvancedMultiple buildings / regions
Introduced at Pilot
Measurement to your zone map
Workplace director release
Sensor-coverage baseline
Introduced at Production
Reporting by zone
Publication workflow in your systems
Approved write-back
Badge-and-sensor integration
Introduced at Advanced
Multi-building estates
Cross-period study packs
Large sensor estates
Multi-floor aggregation controls
Build price From $5,000 From $8,000 Custom quote
Final build priceConfirmed after discovery based on integrations, workflow complexity, estate size, approval controls and deployment requirements.
Separate from buildBuild pricing is separate from recurring Agent Care, which covers managed monitoring, evaluations, incidents and verified improvements after launch.

What we need from you

What you bring, and what we build with it

Each input maps to a piece of build scope and a week in the delivery timeline.

You bringWe build with it
01Your buildings and the zone map behind each one Zone mapping and reading captureWeek 1
02Representative badge, sensor and audit readings Reading binding, method logic and the coverage baselineWeek 2
03Your configured policy on personal data Method mapping, period binding and the automation boundaryWeek 1
04Access to relevant APIs, feeds or exports Badge, sensor and floor-plan source assessment, then integration setupWeek 2
05Studies you would not want questioned Coverage cases and failure-mode testingWeek 4
06What no occupancy figure may prove Coverage scoring, review routing, guardrails and release controlsWeek 3
07A named workplace director to publish Publication workflow, then pilot and production validationWeeks 5–6
Nothing else is required Deployment, documentation and Agent Care handover are ours.

Delivery timeline

Four phases across six weeks

Each bar is sized to the weeks that phase really costs, and the fifth of them therefore doubles up.

Phase W1W2W3W4W5W6
Discovery W1
Build W2 – W3
Evaluate W4 – W5
Pilot & Launch W5 – W6
Week focus W1Workplace discovery, zone mapping and the automation boundary W2Source integration and the sensor-coverage baseline W3Measurement assembly, method logic and release controls W4Evaluation suite, coverage cases and failure-mode testing W5Floor-plan integration, pilot zones and targeted corrections W6One occupancy study run under the workplace director, then Agent Care handover
Reading the bandNo band is padded out; each spans the weeks its own work is named for, and week five holds a pair.
At the end of W6Once the occupancy record validates, Agent Care assumes the agent.
DurationSix-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.

Next step · Facilities AI agent

Build a utilisation agent around the wing your board pack has never actually measured.

Bring us one wing and the readings behind it. Send a floor plan and a single measured week as the next step, and we will return the study with its coverage stated and every unmeasured area named. Whether a floor is released stays with your workplace director.

Nestack Agents · Space utilisationAGT-AF-07 · Agent Care available after launch