Healthcare AI agent · Care coordination
Appointment & Care-Coordination AI Agent
Book, reschedule and sequence visits against your provider templates, chase what each visit needs and close the referral loop — with clinical urgency and every care-plan change kept with a clinician.
- Typical delivery
- 4–6 weeks
- Deployment
- Your stack
- Clinical urgency
- Clinician only
- After launch
- Agent Care
What this agent does
Books the visit, not the priority
In
- 01
Take referrals, orders, care-plan steps and patient requests from the EHR, the referral queue, the portal and phone.
- 02
Read the provider template, visit-type rules, prerequisite list and authorisation status already on file.
Reason
- 03
Match the request to the visit type and provider template the order specifies, at the slot length that visit type carries.
- 04
Sequence a care plan so the imaging, labs, prep and authorisation land before the visit that depends on them.
- 05
Check consent, contact preference, language and transport need against the record before anything is sent or booked.
Decide
- 06
Detect a rule it cannot satisfy, a missing prerequisite or an authorisation not in place, and hold it for a person.
- 07
Detect a request that implies a change of urgency or of the care plan, and route it to the clinician who owns it.
Out
- 08
Book, reschedule, confirm and chase inside the rules, then return the visit outcome and report to the referring clinician.
- 09
Retain the rule applied, the slots offered, the messages sent and every correction a scheduler or clinician makes.
Product statement
The agent books, sequences and chases inside rules a person set. Clinical urgency and any change to a care plan stay with a clinician.
Example workflow
One referral, end to end
- 1Request receivedReferral, order, care-plan step, portal request or a cancellation to backfill
- 2Rules readVisit type, provider template, prerequisite list, authorisation status and contact preference
- 3Slot and sequence proposedOnly slots the template allows, in the order the care plan needs them
- 4Prerequisites checkedPrep, imaging, labs, authorisation, transport and interpreter — each in place or named as missing
No human action required
Stages 1 to 4 run without a person in the loop — the rules are read and the prerequisites checked before anyone is asked to look. A request that touches clinical urgency ends that stretch on the spot.
Books and confirms on the approved path.
Goes to a scheduler, or to the clinician who ordered it.
Scheduler or clinician
Sees the rule that blocked it, the slots that were open and what is still missing, then books, overrides or changes the plan.
Book · Override the rule · Send to the clinician
Booked — handed back
- 6Booked and confirmedWritten to the schedule only where write access and policy allow; urgency is not changed
- 7Loop closed and evaluatedAttendance, prerequisites at the visit, the report back to the referrer and corrections by cohort
Corrections
Slots a scheduler rebooks by hand are counted in the evaluation.
What should not run autonomously
Human approval stays in control
Outside the boundary — human approval required
8 items
- Deciding how soon a patient needs to be seen.
- Reordering a clinical dependency in a care plan.
- Choosing a visit type the order did not specify.
- Double-booking or overriding a provider template.
Automation boundary
Agent acts unaided
- ✓Book, reschedule and confirm inside the provider template and visit-type rules.
- ✓Sequence a care plan so each visit's prerequisites land before it.
- ✓Chase prep, imaging, labs, authorisation, transport and interpreter cover.
- ✓Return the visit outcome to the referring clinician and track what is still open.
Write actions run only inside the approval boundaries agreed during implementation. Clinical urgency is not one of them.
- Booking a service before authorisation is in place.
- Cancelling or moving an appointment a clinician set.
- Sending clinical advice or a result to a patient.
- Changing waitlist, backfill or reminder rules.
Example output
One referral, annotated
Everything the agent proposes is attached to the referral and the rules it was read against.
Coordination output · single referral
Illustrative example- Referral
- Cardiology — referrer marked routine
- Order specifies
- Echo before the clinic visit
- Authorisation
- Not on file
- Proposed sequence
- Echo first, clinic after
- Confidence
- 93%
- Urgency
- As the referrer set it
As received
The referral, the urgency the referrer marked and the order's own sequence, kept as sent — nothing on this side is re-decided.
Evidence used
- Provider template rule
- Visit-type slot length
- Authorisation status
Why it did not book
The sequence fits the template, but the echo has no authorisation on file — so the pair is held.
What the score decides
Confidence decides whether a scheduler checks the slot, not how soon the patient is seen. The referrer sets urgency.
Action
BookOverride the ruleSend to the clinicianValue
Where AI adds value
The same four claims, placed at the point in the workflow where each one applies.
Where the value lands
Value 01 – 04
Every referral and order
Referral queue, orders, portal and phone
Apply the client's own booking rules
Use the provider templates, visit-type rules, prerequisite lists and authorisation status already on file.
Take the chasing off the schedulers
Slots, prerequisites, confirmations and the report back to the referrer are worked without someone holding a list and a phone.
Put the exceptions in front of a person
Rule conflicts, missing authorisation and anything touching urgency reach a person before it is booked.
Build an evidence trail
Retain the rule applied, the slots offered, the prerequisites chased, the messages sent and the scheduler's correction — on both paths.
Integrations
Typical integrations
Five system groups connect to the same agent. Which of them are in scope is decided in discovery.
EHR scheduling & referrals
- Epic Cadence
- Oracle Health
- athenahealth
- Referral and order queues
Templates & visit types
- Provider templates
- visit-type rules
- Slot length
- block and overbook rules
Orders, prep & authorisation
- Imaging and lab orders
- prep lists
- Authorisation status
- transport booking
Agent
Appointment & care coordination
- Reads the rules
- Books and sequences
- Closes the loop
Patient contact
- Patient portal
- secure messaging
- SMS
- Telephony
- interpreter booking
Observability & evaluation
- OpenTelemetry
- Langfuse
- Supported monitoring/evaluation sources
Integration availability depends on the client's existing systems and API access.
Agent controls
Six layers between the model and your schedule
Each control wraps the one inside it. A booking clears every layer before it reaches the schedule, and clinical urgency sits outside all six.
L6 · Outermost — last line of defence
Inward → L1 · closest to the model
- L1 – L2
- Decide whether the slot may be offered
- L3
- Decides what may be sent, and to whom
- L4 – L5
- Keep urgency with a clinician and the trail intact
- L6
- Pulls automation back when signals degrade
How Nestack evaluates it
Evaluate the whole booking — not only whether the slot was filled.
Coverage runs the whole depth of the workflow, and every layer is cut by slice.
Surface — the appointment the patient is offered
- E1Final-output evaluationWas the visit type, provider and slot length correct?
- E2Step-level evaluationDid it read the right template, rule and prerequisite list?
- E3Tool evaluationDid it read and write the correct patient, order and schedule?
- E4Prerequisite completenessWas everything the visit depended on in place on the day?
- E5Slice evaluationHow does booking quality change across specific patient cohorts?
- E6Business outcomeDid the referral loop close, and how much was rebooked by hand?
Floor — the visit that happens as it was planned
Failure modes
Where each failure originates in the agent
Seven failure modes plotted against the five stages of the agent lifecycle.
Agent lifecycle
Direction of processing →
01 · Request / retrieval
1 mode
- AC-01
Prerequisite not in place
The lab, prep or approval is missing on the day.
Stage gathers
The referral, the order, template rules and approval
02 · Rule fit
2 modes
- AC-02
Wrong visit type booked
Booked at a slot length the order does not fit.
- AC-03
Sequence inverted
Imaging lands after the visit that needed it.
Stage proposes
Visit type, provider, slot length and visit order
03 · Booking / write
1 mode
- AC-04
Template rule broken
Blocked time is used or a slot is double-booked.
Stage books
Only where write access and scheduling policy allow it
04 · Chase & contact
2 modes
- AC-05
Reminder to a stale contact
A reassigned number receives another patient's detail.
- AC-06
Referral loop left open
The visit happens and no report reaches the referrer.
Stage sends
Confirmations, reminders, prep and the closing report
05 · Change / Version
1 mode
- AC-07
Backfill narrows access
Released slots go to whoever answers first.
Stage tracks
Model, prompt, template-rule and backfill changes
- Sev-1 · acted outside the boundary
- Sev-2 · the visit is wasted or cancelled
- Sev-3 · the loop stays open or access narrows
Affected slices
A high fill rate says nothing about who got the slot
Aggregate booking quality can look acceptable while a small number of patient cohorts carry most of the rebooking, most of the wasted visits and most of the lost access. Nestack reports performance by slice, not only in total.
Slice performance — reported separately, not only in aggregate
Illustrative example| Slice | Failure rate | Lift | Lift vs. threshold | Status |
|---|---|---|---|---|
| Patients without portal or SMS | 5.6% | 3.5× | Review | |
| Multi-step care-plan sequences | 4.2% | 2.6× | Review | |
| Referral-driven first appointments | 3.1% | 1.9× | Watch | |
| Routine follow-up rebooking | 1.0% | 0.6× | Normal |
Bar: lift vs. routine-rebooking baseline · scale 0–4.0× · tick marks the 2.0× review threshold
2 of 4 slices over threshold
Evidence-linked improvement
Every rule change moves who gets the next slot
A wasted visit is not closed by rebooking one patient. The fix lands in a template rule, a prerequisite or the backfill order — and that changes who is seen.
Improvement cycle · five stages
Switchback — the path turns at Improve and returns at Learn
- 01
Detect
Wasted visits, rebooking or open referrals move in a cohort.
- 02
Diagnose
Traced to the rule read, the visit type, the sequence, the chase or the backfill order.
- 03
Improve
The template rule, prerequisite list or backfill order is re-approved and version-linked.
- 04
Verify
Re-run against held-out bookings from the affected cohort, including the ones that failed.
- 05
Learn
That booking becomes a regression case and the changed rule enters the scheduling runbook.
Learn → Detect
The return edge. A backfill or template change is an access change, so the next cycle re-reads the parity slices, not only the aggregate.
Typical build scope
Twelve workstreams across six weeks
The build scope read against the delivery timeline. Week structure follows the six-week plan — discovery, rules and templates, sequencing and chasing, evaluation, integration, then production validation and handover.
Workstream
Week 1Week 2Week 3Week 4Week 5Week 6
- 01Workflow discovery and automation-boundary definition.Week 1
- 02EHR scheduling and referral API assessment.Week 1, Week 2
- 03Provider templates and visit-type rules.Week 1, Week 2
- 04Prerequisite and authorisation checklists.Week 2
- 05Referral and order intake, then slot proposal.Week 2, Week 3
- 06Care-plan sequencing and dependency rules.Week 3
- 07Waitlist and cancellation-backfill order.Week 3, Week 4
- 08Consent, contact preference and reminders.Week 4
- 09Transport and interpreter booking.Week 4, Week 5
- 10Evaluation suite, parity slices and regression cases.Week 4, Week 5
- 11Scheduling write-back and loop closure.Week 5, Week 6
- 12Observability, deployment and Agent Care handover.Week 6
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 clinic or service line | ProductionProduction EHR integration | AdvancedMulti-site / multi-service |
|---|---|---|---|
| Introduced at Pilot | |||
| Rule-checked booking and rescheduling | ✓ | ✓ | ✓ |
| Provider template and visit-type rules | ✓ | ✓ | ✓ |
| Clinician-owned urgency and care plan | ✓ | ✓ | ✓ |
| Baseline evaluation | ✓ | ✓ | ✓ |
| Introduced at Production | |||
| Prerequisite and authorisation chasing | — | ✓ | ✓ |
| Care-plan sequencing | — | ✓ | ✓ |
| Reminders and confirmations | — | ✓ | ✓ |
| Referral-loop closure and report return | — | ✓ | ✓ |
| Observability and evaluation | — | ✓ | ✓ |
| Introduced at Advanced | |||
| Backfill with access-parity monitoring | — | — | ✓ |
| Transport, interpreter and site controls | — | — | ✓ |
| Build price | From $5,000 | From $8,000 | Custom quote |
- Final build price
- Confirmed after discovery based on scheduling and referral integrations, template complexity, patient volume, contact channels, approval controls and deployment requirements.
- Separate from build
- Build 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 bring
We build with it
01Your templates, visit-type rules and backfill policy
→Provider templates, visit-type and backfill rules
Week 102The prerequisites and authorisation each visit type needs
→Prerequisite and authorisation checklists
Week 203Access to EHR scheduling, referral and order APIs
→Epic Cadence, Oracle Health and referral-queue integration
Week 204Your care pathways and the dependencies inside them
→Care-plan sequencing and dependency rules
Week 305Consent, contact preference, message and retention rules
→Consent, contact-preference and reminder rules
Week 406Bookings that went wrong, including day-of cancellations
→Evaluation suite, parity slices and regression cases
Week 407Named schedulers and the clinicians who own urgency
→Review workflow, then pilot booking and production validation
Weeks 5–6
Nothing else is required
Deployment, documentation and Agent Care handover are ours.
Delivery timeline
Four phases across six weeks
Phases are drawn over the weeks they actually occupy. Week 5 carries both the parity evaluation and the first supervised bookings.
Phase
W1
W2
W3
W4
W5
W6
- DiscoveryW1
- BuildW2 – W3
- EvaluateW4 – W5
- Pilot & LaunchW5 – W6
Week focus
W1Template rules, backfill policy and where a person has to book
W2Referral and order intake, prerequisite checklists, first bookings
W3Care-plan sequencing, waitlist and cancellation-backfill order
W4Evaluation suite, parity slices and contact-preference rules
W5Scheduling write-back, supervised booking and targeted corrections
W6A referral cycle booked beside the existing process, then handover
- Reading the band
- Discovery closes when the template rules are signed, not when week 1 does. The first bookings in week 5 test the evaluation suite.
- At the end of W6
- Booking has run alongside the existing process and the loop has been closed on real referrals, then Agent Care takes over monitoring.
- Duration
- Six-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.
Next step · Healthcare AI agent
Build a scheduling agent around your access rules.
Show us your provider templates, your referral and order queues, and what has to be in place before a visit goes ahead. We'll book one referral end to end on your rules, then agree what a scheduler clears.
- Typical delivery
- 4–6 weeks
- Deployment
- Your EHR and scheduling stack
- Clinical urgency
- Clinician, always
- After launch
- Agent Care available
Nestack Agents · Appointment & care coordination
AGT-HC-05 · Agent Care available after launch