Healthcare Solutions | Call Center Operations

Premium Products

24/7 Product Support

Operating guideHealthcare Solutions workspace

⌕

5 guides/ search

GET ORIENTED

Start here

The safe daily operating sequence for every Healthcare Solutions operator.

5Five-minute startupComplete these checks before handling customers or releasing outreach.

Confirm the environmentVerify the top badge says STAGING or LIVE.
Check platform statusOpen the lower-left status card and confirm required services.
Sign inUse Secure login before viewing customer records or changing delivery state.
Review exceptionsClear compliance blocks and failed deliveries before new work.
Open the floorReturn to Command Center and select the first active conversation.

MCP & REHEARSALHow can the assistant guide an operational test?+

Open Client Rehearsal → Test Coach. Inspect readiness, use fictional caller details, rehearse one turn, compare measurements, save feedback and reopen the same run. The coach distinguishes browser-send and AI input-end timing, unknown values, delivery and human hearing. Connected voice execution requires the canonical account connection, public tools and an exact plan approved in the signed-in console; a guide or historical receipt does not prove those are enabled. Changing an approved plan needs a new review.

HUMAN WORKFORCEHow do employees become available for handoff?+

Administrators invite individual employees and verify their actual state/product licenses, training, queues and appointments. Employees sign in to their own desk and choose Available. A current session and fresh heartbeat are required; login alone does not prove readiness. Inspect Human handoff readiness before sending an offer. A fictional rehearsal proves offer acceptance only, not a browser media bridge or phone connection. Use Complete rehearsal offer afterward. Embedded dialer tests remain offline until carrier certification and release approval.

STARTUP · 5 MINUTESHow do I start an operator shift safely?+

Use the same sequence every time so an old exception or disconnected provider does not become a customer-facing failure.

Read the environment badge and platform status.
Sign in and refresh the current dashboard.
Review Compliance for blocked or review-required work.
Review Execution Plane for failed, delayed, or dead-lettered jobs.
Open Command Center and begin with the highest-priority active conversation.

NAVIGATIONWhat does every dashboard do?+

Command Center: Live cross-channel work and operator controls.
Receptionist: Inbound intake, disclosure, product routing, appointments, and handoff.
Leads: Customer cards, outreach actions, scoring, and pipeline stages.
Campaigns: Audience, policy, approval, and release preparation.
Execution Plane: Worker, queue, job, delivery, and failure operations.
Scale Lab: Safe volume models, queue benchmarks, lane ceilings, and circuit breakers.
Analytics: Conversion, state, source, channel, and Overwatch performance.
Compliance: Preflight decisions, evidence, reason codes, and audit exports.
Integrations: Twilio, Resend, Google Calendar, DocuSign, and runtime readiness.
EILA Overwatch: Your private text, voice, video, memory, and operational assistant.

ENVIRONMENTHow do I know whether traffic is real?+

The badge at the top of every screen is the source of truth. STAGING · PROVIDERS BLOCKED means the master delivery switch is off. STAGING · LIVE DELIVERY ENABLED means real provider actions can run after their channel and compliance checks. Sign in to read the current persisted setting. Queue bench messages are synthetic even though they travel through real Healthcare-owned queues. Enabling live delivery requires administrator access and an explicit confirmation. Use Stop all provider delivery to block new provider actions; calls or requests already in progress must be ended separately.

ACCESSWhat can I see before and after signing in?+

Sign in to view operational dashboards and customer records. Principals have the owner workspace; licensed agents see only their own agent desk and offers. Administrators manage invitations, credentials, providers, campaigns, evidence exports, and Scale Lab controls.

DATA FRESHNESSWhen does the dashboard update?+

Live floor views refresh on their operating interval; other screens refresh when opened. Use the circular refresh control in the top bar after an external provider event, handoff, or integration authorization. If totals and cards disagree after refresh, use the troubleshooting guide instead of repeating the action.

COMMAND CENTERHow do I read the live conversation floor?+

The KPI row summarizes today. Product colors identify what each customer needs; stage and channel remain secondary tags. Select a conversation card to open the same product color in the inspector. Confirm the customer, source, channel, lead score, compliance state, and live transcript before using an operator control.

COMMAND CENTERWhat do the operator controls do?+

Listen: Monitor the active conversation without speaking.
Whisper: Coach the assistant privately.
Instruct: Send a bounded next-step instruction.
Take over: Move control to a human operator.
Transfer: Route to the approved licensed queue.
SMS / Email: Send the approved follow-up through its dedicated worker.
Disposition: Record the outcome and next action.
End: Close the active session after its disposition is captured.

RECEPTIONISTHow do I handle an inbound caller?+

Select the caller card and verify name, phone, state, and product intent.
Confirm the AI/recording disclosure is accepted before sensitive intake.
Review the product route and customer need in the case file.
Qualify only within the approved script and routing rules.
Schedule the callback or request a licensed-agent handoff.
Confirm the timeline records the disposition and next action.

VISUAL MAPWhat do the product colors mean?+

The ring and tinted background follow the product everywhere: blue Medicare Advantage, red Medicare Supplement, purple Part D, teal ACA Marketplace, amber Final Expense, green Dental & Vision, indigo Hospital Indemnity, and rose Cancer / Heart / Stroke. Workflow stage is a separate tag; it never replaces product identity.

HANDOFFWhen should I escalate to a licensed agent?+

Escalate when the workflow reaches plan-specific advice, enrollment, an explicit customer request for a person, unresolved identity or consent, a script boundary, or any policy reason that requires human review. Confirm the receiving queue and preserve the product, state, transcript, disclosure, and intent in the handoff packet.

LEAD INTAKEHow do I add the insurance intake to a website?+

Open Lead Intake below Overwatch. Edit the copy and review Widget, Page, and Mobile previews. Add the exact HTTPS website origins under Website & collection settings. Select Accept lead requests and save when ready. Set a placement label and copy the standalone link or iframe code. Preview submissions are never saved. Public submissions appear in Leads; open View coverage intake & consent to review the original answers and contact permission. No call or message is sent automatically.

LEADSHow do I read a customer card?+

The colored ring is the requested product. The score summarizes current lead quality; it is not a compliance decision. Read source, campaign, consent, last contact, state, and customer need before contacting. Use the call, text, email, and callback buttons only after verifying the channel is allowed.

LEADSWhich pipeline stage should I choose?+

New lead: No verified outreach has been completed.
Contacted: A real two-way or documented outreach attempt exists.
Qualified: Need, state, channel consent, and routing requirements are present.
Nurture: Valid future opportunity that is not ready now.
Booked: A confirmed callback or licensed-agent appointment exists.
Won / Enrolled: The organization-approved success outcome is recorded.
Lost: Closed with a recorded reason; do not use this to hide blocked work.

LEADSHow do I find and prioritize customers?+

Search by name, phone, state, product, or source. Use product chips to isolate the requested insurance line and stage chips to isolate workflow state. Start with active callbacks and qualified customers, then high-intent new leads, then scheduled nurture work. Compliance blocks always take priority over speed.

LEADSWhat must be present before importing a lead?+

Require a usable identity, destination, state, requested product, source, acquisition timestamp, channel-specific consent evidence, and suppression/DNC status. Preserve the original attribution. Reject or quarantine duplicates, missing consent, impossible destinations, and rows whose jurisdiction cannot be determined.

CAMPAIGNSHow do I build a campaign?+

Create a draft and select one product objective.
Choose the approved channel and audience source.
Attach the reviewed script, disclosure, and offer version.
Set jurisdiction, local-time window, pacing, and handoff queue.
Run compliance preflight and resolve every block.
Freeze the policy dimensions, obtain human approval, and only then release.

CAMPAIGNSWhy can’t I release a campaign?+

Release remains unavailable when the campaign is a draft, the audience lacks consent attribution, a policy dimension changed after review, the approved script is missing, the sending provider is not ready, or a compliance decision is BLOCK / REVIEW REQUIRED. Fix the evidence and repeat review; never work around the gate.

CAMPAIGNSHow do I pause outbound work?+

Pause the campaign to stop new orchestration, then open the relevant lane circuit in Scale Lab if jobs must stop entering a channel queue. Existing in-flight provider work may still report final status. Record the reason, watch delivery reconciliation, and resume only after the underlying condition is cleared.

HIGH VOLUMEHow should a million-call day be prepared?+

Model first, then prove a bounded synthetic queue bench, then release in controlled cohorts. Segment by state/local time and product, keep dedicated voice/SMS/email/video lanes isolated, verify provider and licensed-agent capacity, define stop thresholds, and monitor connect rate, complaints, delivery lag, and compliance exceptions throughout the release.

EXECUTION PLANEHow do I read worker and queue health?+

Follow a job from ingress through the governor into its dedicated channel queue and worker. Compare accepted, queued, active, completed, deferred, failed, and dead-letter counts. A healthy ingress with a growing queue indicates downstream capacity pressure; a flat queue with provider failures indicates delivery or credential trouble.

EXECUTION PLANEWhen should I retry failed work?+

Retry only after identifying whether the failure is transient, permanent, provider-side, destination-side, or compliance-related. Never retry DNC, invalid consent, forbidden window, or permanent destination errors. Preserve the original job and provider IDs so delivery-status reconciliation and audit history remain intact.

SCALE LABHow do I model one million jobs safely?+

Select the voice, SMS, email, or video lane.
Enter 1,000,000 jobs and the intended operating window.
Review required per-minute throughput against the lane ceiling and daily capacity.
Adjust burst and shards only within tested infrastructure limits.
Save the model for comparison; the model creates no queue messages and contacts nobody.

SCALE LABHow do I run a real-path queue bench?+

Use Hardware queue bench only after the dry model passes. Choose a bounded 100–1,000 job batch, confirm the screen says synthetic/provider bypass, run one lane at a time, and watch every job reach a terminal state. Stop if dead letters, unexplained lag, or provider traffic appears.

SCALE LABWhat do circuit breakers do?+

Open circuit defers new work for one lane without deleting it. Open all circuits is the emergency stop across channels. After the root cause is documented and cleared, resume individual lanes first, verify terminal outcomes, and only then restore the remaining lanes.

ANALYTICSHow do I read the conversion funnel?+

Read attempts → connected → qualified → booked/completed as separate stages. Compare both counts and rates; a large top-of-funnel increase can hide lower quality. Filter by product, state, source, channel, campaign, and time window before deciding where the process is failing.

ANALYTICSHow do I find a weak state, source, or product?+

Hold the time range and funnel definition constant, then compare one dimension at a time. Check volume before trusting a rate. Open the underlying customer cohort to confirm consent, source attribution, product intent, dispositions, and compliance decisions explain the aggregate result.

OVERWATCHWhich alerts need attention first?+

Prioritize active compliance exposure, customer-facing delivery failure, stuck licensed-agent handoffs, growing dead-letter queues, then stale work. Open the source dashboard for every alert, confirm the affected cohort, and record the operator action rather than clearing the symptom alone.

COMPLIANCEHow do I run a preflight decision?+

Choose state, product, channel, and inbound/outbound direction.
Enter customer-local time and channel-specific consent evidence.
Confirm suppression/DNC, licensing, appointment, training, disclosure, recording, and script evidence.
Run preflight and read every reason code.
Proceed only on ALLOW; BLOCK stops work and REVIEW REQUIRED routes to an authorized human.

NON-NEGOTIABLEWhat do I do with a DNC or opt-out?+

Stop the affected channel, record the request with its source and timestamp, update the suppression record, and do not retry or move it to another channel as a workaround. If scope is unclear, fail closed and route to compliance review.

CALL SAFETYWhat must happen before sensitive intake?+

Use the currently approved AI/recording disclosure for that state, direction, and channel; capture the acceptance evidence; stay inside the approved script; and transfer plan-specific advice or enrollment to the licensed workflow. Missing or ambiguous evidence is REVIEW REQUIRED, not implied permission.

AUDITHow do I preserve or export a decision?+

Open the decision and confirm the customer/job reference, policy fingerprint, inputs, evidence references, reason codes, actor, and timestamps. Authorized administrators can export the record. Operational software enforces the organization-approved policy set; it does not replace counsel or regulator guidance.

TWILIOHow do I attach the Healthcare phone number?+

Open Integrations and validate the bound account credential.
Discover and choose a Healthcare Solutions-owned number.
Confirm the displayed inbound and status callback URLs.
Run validation with live delivery disabled.
Complete the explicit production activation gate only after ownership and policy approval.

RESENDHow do I change the email provider or sender?+

Sign in as an administrator, open Secure login → Providers, choose Resend or Disabled, and update the sender name, verified sender address, and reply-to. Saving changes routing metadata only; secret values stay in Cloudflare Secrets Store and are never displayed or copied into the dashboard.

GOOGLE CALENDARHow do I connect and publish appointments?+

Authorize Google Calendar from Integrations, finish Google consent, return to the callback, and use Verify. In the receptionist case, schedule the callback, verify date/time/time zone and customer details, then publish. Refresh the case and confirm an external event reference was recorded.

DOCUSIGNHow do I send and track a document?+

Authorize and verify DocuSign, save the reviewed template ID and exact recipient role, then open a consented case with a valid email and selected product. Send the template once, record the envelope ID, and refresh its status from the case file. Do not duplicate-send because the status looks delayed.

CREDENTIALSWhere are provider secrets managed?+

The application consumes account-level secrets only through named Cloudflare bindings. This UI shows readiness and routing metadata, never secret values. Missing bindings must be corrected through the authorized platform deployment process; do not paste keys into customer records, source code, support chat, or provider settings.

EILA ASSISTANTHow do I use EILA text support?+

Sign in, open EILA Overwatch, and ask a concrete operational question: name the dashboard, customer or campaign, symptom, and expected outcome. EILA can summarize current aggregate facts and guide the next operator action; it cannot silently approve policy or release live traffic.

EILA ASSISTANTHow do voice and video modes work?+

Browser call uses this device’s microphone and spoken playback. Video creates a private LiveKit room through the committed Healthcare tenant adapter. Confirm the readiness grid before starting. If a room opens but the avatar does not join across multiple tenants, stop—this is a shared-runtime incident, not a reason to patch this product.

PERSONALIZATIONHow do Jake or Matt configure their assistant?+

Sign in, open Secure login → Profile, then set email, phone, assistant name, personality preset or custom instructions, and greeting. Upload a PNG/JPEG/WebP portrait and an optional WAV reference. Both owners share the same runtime, while profile assets and settings remain partitioned by authenticated user.

PRIVATE MEMORYWhat does EILA remember?+

Conversation history and explicit memories are stored inside the signed-in user partition. Save durable preferences, goals, workflows, or identity context with clear labels. Review and delete outdated memories from the assistant panel. Jake’s content is not Matt’s content even though they use one shared runtime.

QUICK RECOVERYThe dashboard looks stale or frozen—what now?+

Use the top refresh button once.
Confirm the platform status card is operational.
Check that search and filters are not hiding records.
Open Execution Plane for delayed or failed jobs.
If totals still conflict, capture the dashboard name, time, filters, and affected customer/job before asking EILA.

COMMAND CENTERWhy is the live floor empty?+

Confirm the channel/status filters, environment, and time window. Completed calls may have aged out of the live floor and remain in receptionist/customer history. Then inspect ingress and queue counts in Execution Plane. Do not create fake provider traffic to make the floor appear populated.

INTEGRATIONSA provider action failed—should I press it again?+

First read the recorded status, error class, destination, and provider reference. Verify readiness in Integrations and reconciliation in Execution Plane. Retry only a confirmed transient failure and only once through the approved retry path; never duplicate a call, message, calendar event, or DocuSign envelope because a callback is delayed.

COMPLIANCEMy campaign is BLOCKED or REVIEW REQUIRED+

Open the preflight record and resolve each reason code at its source: consent, DNC, local time, licensing, training, disclosure, recording, script, provider, or frozen policy dimensions. Re-run preflight after the evidence changes. REVIEW REQUIRED needs the designated human reviewer; BLOCK is never an operator override.

EMERGENCYHow do I stop unexpected outbound volume?+

Open Platform & Providers and select Stop provider delivery, then open all circuits in Scale Lab.
Pause active campaign releases.
Confirm queued work is deferred and no new orchestration is entering lanes.
Watch provider delivery reconciliation for in-flight work.
Record the time, actor, affected lanes, campaign, and reason.
Resume one lane at a time only after authorized review.

SECURE WORKSPACE

Account & settings

Protected operator accessSign in to run campaigns, control conversations, export evidence, and use the shared executive assistant.

Delivery policy & sender defaults