Existing customers

AI customer service for the questions you already answer the same way.

Support is a different job from winning the next lead. EXCEL’s customer-service agent is built for people who already know you: hours, policies, how to submit a request, and when a person has to step in.

It answers from approved services, hours, and policies. It does not invent a refund, give professional advice, or handle disputes on its own.

Elliot is a live receptionist demo. Use it to hear conversational EXCEL AI — then we design support around your actual policies.

What a support agent is for

If the same five questions burn the same staff hours every week, that is the work. Unusual complaints stay with people.

  • Repeat policy and how-to questions

    Hours, cancellation wording you approve, how to submit a work order, where to send documents.

  • Status requests, with a hard stop

    If a secure lookup is in scope and connected, the agent can retrieve what you allow. Otherwise it captures the question and routes it. It should not guess that a part shipped.

  • Ticket capture

    When the answer is not on the approved list, the agent still leaves a usable record instead of a voicemail fragment.

  • Phone or messaging, where configured

    The same knowledge can sit on the phone line, a text thread, or both — depending on the channels in the build.

Why support still sits behind new leads

One number, no triage

Existing customers wait behind sales callers for answers that are already on the website.

Inconsistent answers across the team

Whoever picked up quoted a different cancellation window. That becomes a complaint later.

Every contact treated as a ticket that needs a human

Including ‘what time do you open?’ Automation should shrink tickets, not hide them.

A support conversation, by design

Identify, answer or capture, escalate. The agent is not a personality. It is a first pass.

  1. 01

    Identify

    New versus existing, and which account, unit, or job if you collect that.

  2. 02

    Answer from the list

    Only published or approved information. No improvising a policy.

  3. 03

    Capture the rest

    A structured request: who, what, when, and how to reach them.

  4. 04

    Escalate exceptions

    Billing, complaints, clinical, legal, or anything you marked human-only.

Hypothetical: property-management work order

Illustrative tenant support flow. Not medical, legal, or a claim about a specific client.

Tenant
The AC in 2B hasn’t cooled since yesterday. I already emailed last week.
EXCEL AI
I can start a maintenance request. What’s the unit again, and is anyone in the home who needs cooling for a medical reason — I will note that for the team, not give medical advice.
Tenant
2B, and yes, my father is here. Can you send someone today?
Then
The agent collects access notes and follows the property’s after-hours vendor rules. It does not promise a same-day tech unless that promise is an approved rule.

Support agent versus receptionist

They can share knowledge. They should not share the same job description.

Receptionist is the front door

New work, new leads, first contact. Support is for people already in the system.

Support refuses to guess

A good support agent that does not know an account status asks, captures, and routes. A bad one hallucinates a refund.

Tickets should get smaller

If every chat still becomes a human thread, the approved-answer set is too thin — that is a content problem, not a model problem.

What gets configured

  • The exact policies the agent may quote
  • What it must never discuss (pricing exceptions, clinical, legal, HR)
  • Whether any account lookup is in scope
  • The inbox, help desk, or email the ticket lands in
  • Phone, SMS, or both

Human-only by default

Disputes, refunds, threats, HIPAA-sensitive details unless a system is formally designed for them, and anything missing from the approved set. Capture and route. Never advice.

Calendar booking works when a calendar or scheduling system can be securely connected. CRM, SMS, and other software connections are scoped during discovery: some use supported APIs or webhooks; others are custom-built. EXCEL does not claim every platform is pre-integrated.

Plan details live on the pricing page. Current plans are listed there.

Industry overviews

Industry overviews live on their own pages. Dedicated workflows, when they exist, are linked below.

Teams that live on repeat questions

A few industries where this service actually behaves differently — not a directory of every pair.

Guides related to this service

Related guides for the problems this service is built to solve.

Related EXCEL services

Questions buyers actually ask

Can it access our customer records?

Only if a secure lookup is in scope and connected. Many first builds answer public or approved information and capture tickets for the rest.

What if it does not know the answer?

It should ask a clarifying question, capture the request, or escalate — not invent a policy.

Can it handle billing disputes?

Those should be marked human-only. The agent can collect context and route the conversation.

Is this the same as the receptionist?

No. The receptionist is inbound front-desk work. Customer service is for existing-customer questions, status, and tickets. They often share knowledge.

HIPAA, legal, or medical questions?

Out of scope unless a system is formally designed for them. Default is capture and route, never advice, never a claim of certification that is not in place.

Can it sit on SMS as well as the phone?

Messaging can be part of the same knowledge set when a messaging path is connected. SMS follow-up is a Professional-and-above capability in typical EXCEL plans.

Put the easy answers on the front of the conversation. Keep people for the ones that are not easy.

Bring the real call types, the real software, and the rules you already use. We will tell you what belongs in an agent and what should stay with people.