An AI virtual assistant for the LCHS Hub
Picks up the nuisance issues that swamp the Hub — password resets, account lockouts, forgotten Teams diversions, software installs, frozen computers — and actions what it can. Logs every interaction to Jitbit. Skill-matches the escalations. Frees the Hub team for the work that needs them.
Try the LCHS Hub assistant
Open the chat below. Try any of the prompts — the AI looks up the staff member, actions what it can (Azure AD reset, Teams call forwarding, Intune install), and creates a rich Jitbit ticket either way.
Click any prompt to copy it.
What the assistant handles
Five action-led workflows covering the nuisance volume Mark flagged, plus a knowledge fallback. Every interaction creates a Jitbit ticket for the audit trail — even fully-resolved ones.
Password reset
Verifies staff against Azure AD, sends reset link to the email on file (not a new one they offer — security boundary), logs Jitbit ticket. Actual reset, not a deflection to "log it on Jitbit yourself".
Account unlock
Locked from too many attempts? Unlocks Azure AD on the spot. Chains a password reset if they've forgotten that too. Distinguishes locked from disabled (HR-managed) and escalates the right path.
Teams phone diversion
Forgot to divert before a meeting? Set forwarding to a colleague, group (e.g. Allied Health Reception), or voicemail. Reads back the rule + duration before firing — wrong target is a real consequence.
Software install via Intune
Checks the staff member's Intune catalogue entitlement. If they're entitled (Power BI, Adobe, Zoom, etc.), pushes the install — ~13 min to ready. Outside the catalogue → logs a Hub-review ticket instead. Won't pretend it can install what it can't.
Computer / app troubleshoot
Walks staff through proper restart, app reset, VPN reconnect. Detects hardware-shaped issues (won't power on, blue screen) and escalates straight to L2 Endpoints with full context.
Urgent clinician escalation
Clinician mid-session, family violence advocate with a client, time-critical case — detects the urgency from the opener and routes to Hub priority queue immediately. No troubleshooting walk-through, no SLA waiting. Hub picks up within 90 seconds.
Voice channel — same assistant, on the Hub line
Same workflows, same actions, same skill-matched escalation — on voice. Emma AU voice, English STT tuned for LCHS terminology (Jitbit, Touchpoint, Intune, Best Practice, MediCS, allied health). Phone number's being provisioned for your test; we'll attach it before our follow-up.
How we built this
Wired up specifically for LCHS's stack and the nuisance categories Mark identified.
Modelled your stack
Mapped Jitbit (ticketing), Azure AD / Entra (identity), Intune (endpoint), Teams (telephony), SharePoint (KB), Touchpoint (Hub queue). The business context primes the AI on LCHS's real flow.
Built the workflows
Six topic-matched workflows plus a router. Each looks up the staff member first, then actions or walks through. Urgent clinician-with-client detection in the router skips troubleshooting entirely.
Mocked your APIs
Nine mock tools shaped like Jitbit REST, Azure AD Graph, Teams Graph, and Intune. Lookups + actions (reset, unlock, forward, install, ticket-create, escalate). Swap mocks for your real APIs in production — Mark confirmed Jitbit's API is strong, integration would be straightforward.
Identity-free for testing
The mock data is generic — type any LCHS email or name in the chat, the AI rolls with it. Lets you and the team test scenarios without setting up fake users.
What's next — POC to production
Four steps from this demo to a live deflection layer on the Hub line.
Train on your SharePoint guides
Point Lorikeet at the ~90 SharePoint guides you mentioned. The AI grounds every general how-to answer in your actual content, not approximations.
Phase 1 — read-only on real APIs
Replace mock lookups with real Jitbit / Azure AD / Intune API calls. AI answers and walks through using actual staff data. Zero write actions yet — purely deflection upgrade on the Hub line.
Phase 2 — write-back, one action at a time
Enable each action with confidence gates — password reset first, then unlock, then phone forwarding, then catalogue installs. The AI earns each action via measured success rate.
Touchpoint integration + omnichannel
Route Hub handovers into the right Touchpoint queue with full context. Once the IT helpdesk POC is working, the same agent + orchestration extends to the long-term client-facing omnichannel vision (booking, service info).
Built for LCHS's reality
Healthcare-aware, urgency-aware, action-led. Designed to deflect ~600 nuisance tickets/month without compromising the urgent clinical path.
AU data residency
Multi-region. LCHS staff and client data can stay in AU. Important for community health privacy posture.
Action-led, not deflection
The agent doesn't say "log a ticket on Jitbit yourself" — it resets the password, sets the forwarding, pushes the install. That's the gap vs answer-only tools.
Urgent clinical override
Clinician with a client, family violence advocate with a client, time-critical clinical case — instant Hub priority routing, no troubleshooting, 90-second response. Demo prompts include this scenario.
Skill-matched Hub escalation
L1 Helpdesk, L2 Endpoints, Identity & Security, Telephony, Clinical Apps, Network. When the AI hands off, the human picks up with full context and the right specialism, not generic.
Phased API permissions
Start read-only against Jitbit + Azure AD. Enable write-back per action as confidence builds. The AI only sees and actions what you've authorised — fine-grained, auditable.
Coach — 100% post-hoc review
Every conversation reviewed against quality criteria. Patterns surface — what's resolving cleanly, where the AI's escalating, where SharePoint guides are thin. Self-improving loop.
Ready for the POC?
Happy to walk through the executive summary success metrics and what Phase 1 deployment against your real Jitbit + Azure AD looks like.
Talk to the team