.svg){kind=small}
Resolve rider tickets
with ride context.
Levy drafts support replies using the rider's live ride, payment history, vehicle state, wallet balance, and your fleet policies. Common issues get handled fast; risky ones stay with a human.
Draft reply ready
Battery died during active ride
Rider message
My scooter shut off before I could end the ride. I'm still being charged and can't lock it.
I found your ride and confirmed the scooter is locked. I've stopped billing at the last GPS movement and credited the extra $1.72 back to your wallet.
How It Works
The ticket does not start blank.
The support draft is generated from operational truth, not a generic chatbot answer. That is the difference between apologizing and actually fixing the rider's issue.
Ticket arrives
A rider submits a mobile ticket with subject, category, language, phone, and message. Pre-auth riders can still be matched by phone.
Context loads
Levy attaches customer profile, active ride, recent rides, vehicle number, payment state, wallet balances, and subaccount details.
AI drafts safely
The model prepares a reply with confidence, policy grounding, and clear next steps. Refunds and liability-sensitive calls stay approval-gated.
Agent approves or escalates
Support can send, edit, request more detail, or escalate to operations with the full audit trail already attached.
Support Automation
Fast where it should be. Careful where it has to be.
The goal is not to replace judgment. It is to remove repetitive lookup work so support can spend time on the tickets that actually need a person.
No silent liability admissions
Accidents, injuries, insurance, damage, and police reports route to human review with prepared context.
Refunds stay approval-gated
The draft can calculate a proposed credit, but the actual wallet or Stripe action remains explicit.
Language-aware replies
Ticket language is detected and preserved so support can answer riders without opening a separate translator.
Human handoff by design
Low confidence, missing context, repeated rider frustration, and policy exceptions go straight to an agent.
Ticket Types
Built for the weird edge cases riders actually send.
Generic support bots do not know whether the vehicle is locked, whether a charge captured, or whether the rider has an active ride. Levy's support AI starts from the fleet record.
Configure Support AICan't unlock or lock
Check active ride state, command history, vehicle online status, and nearby alternatives.
Billing confusion
Explain unlock, time, distance, parking, promo, wallet, card capture, and refund status.
Ride would not end
Find last movement, stop overbilling, and queue a field check if the vehicle is still unlocked.
Reservation or rental help
Pull the booking, location, model, package, and next allowed action into one response.
Give every agent the whole ride story.
Turn support from manual investigation into review, approval, and exception handling.
Proof & readiness
Connected to the support stack you already run.
AI rider support is useful because it starts with the real ride, vehicle, payment, wallet, and customer record instead of a generic help article.
Grounded in
What ships
- Support ticket context builder
- AI draft response generator
- Confidence and language detection
- Human approval workflow
- Ticket audit trail
Deployment timeline
- 1
Day 1 — Policy import
Support guide, refund rules, and escalation rules loaded.
- 2
Day 2 — CRM wiring
Ticket source and draft endpoint connected to your support flow.
- 3
Day 3 — Agent review
Support team reviews canned drafts and guardrails.
- 4
Day 5 — Live
AI drafts appear for real tickets with human approval.
Controls included
- No autonomous refunds or liability admissions
- Low-confidence human handoff
- Subaccount-level policy grounding
- Prompt and draft audit logging
Keep exploring
Related products
Add AI rider support
Tell us how support works today. We'll show where AI can draft, resolve, and escalate safely.