HighLevel AI Employee
HighLevel AI Employee setup: chat, voice and the knowledge base behind them
GoHighLevel (GHL) groups its AI products under one name, the AI Employee. We set up the parts that talk to your customers, chat and voice, and we write the knowledge base they answer from before either one goes live. Rob Eason, a HighLevel Certified Admin, plans it with you, and you approve every word first.
What we set up
Three parts, built in the right order
The knowledge base comes first, because chat and voice both answer from it.
- A sorting pass before any writing: each agent is marked chat or voice, then single prompt or flow, so nothing gets built in the wrong builder.
- A knowledge base in six parts: the business, pricing and programs (or a line that sends price questions to a call), what working with you is like, how you operate, rules for the AI, and FAQs as clean question and answer pairs.
- FAQs that always cover the next step, where you serve, what happens to price questions and what a person can expect after they book.
- Knowledge seeded from your website. A sales call recording can confirm a detail, but a transcript is never pasted in as an answer.
- A written knowledge base instead of a website sync when your public site says things your business no longer does.
- Chat agents with the knowledge base attached to the agent rather than pasted into the prompt, which keeps the prompt to about a kilobyte and easy to check line by line.
- Voice agents on HighLevel's own voice builder by default, with the inbound and outbound sides set up apart.
- A named person for every handoff the knowledge base cannot cover.
How a setup runs
From a call with Rob to agents you have tested
A 1:1 with Rob
Thirty minutes on what the AI should handle, which channels it covers, and who on your team takes over when it should stop. Rob opens your HighLevel account on the call instead of talking in general terms.
One plan, every word in it
You get a single document with each agent's prompt, each opening line and each knowledge base entry written out in full. Nothing is built until you have read it and said yes.
Build, then try to break it
We build in your account and test with chats and calls meant to trip it up: a price question, an off-topic question, someone who asks for a person straight away.
Handover with proof
Your QA report shows each agent and the knowledge base working, with screenshots, so you can see exactly what was tested and what the AI said back.
Changes after launch
Before we touch anything you ask to change, you hear whether it is a fix inside what you approved or new scope. The work starts after that.
What you bring
What the build needs from you
Your website, and where it is wrong
The knowledge base starts from your site. Tell us which pages are out of date and what you will not put in writing, such as prices.
The people who take over
A name and a number for each kind of handoff: a sales question, an existing customer, a complaint, an urgent call.
Your calendar rules
The calendar the AI books into, and any days, people or appointment types it should leave alone.
The tool your jobs live in
If a field-service tool or practice system owns scheduling, the AI collects what that tool needs before the contact syncs. For an artificial turf installer, that meant a first name, a phone or email, and the job address.
Where it stops
The AI answers from the approved knowledge base and nothing else. It does not quote a price you have not written in, give medical, legal or financial advice, or promise outcomes. It is not a replacement for your staff: it covers the hours and channels named in the plan and hands everything else to a person.
By business
Where this gets set up
Insurance agencies
HighLevel for Medicare and other insurance agencies: a pipeline that follows each client past the policy, consult booking, and AI that answers in two languages.
See Insurance agenciesReal estate agents and brokerages
The first text to a web lead, a nurture for the ones who aren't ready, and an AI line that catches the calls a broker misses.
See Real estate agentsArtificial turf installers
Inquiry forms that lead to a booked estimate, follow-up for anyone who doesn't book, and new contacts passed to Jobber, with AI that gathers details for your team.
See Artificial turf installersPersonal injury law firms
An intake pipeline from the first inquiry to the signed client, with AI answering and consultation booking, for injury and property damage firms.
See Personal injury law firmsFertility clinics
Free advisory calls booked from your calendar, paid consultations sent a booking link that takes payment, and clinical questions passed to a person.
See Fertility clinicsPharmacies
A Voice AI trainer that plays a prescriber, so your sales reps can rehearse office calls without a single real patient record in the AI.
See PharmaciesLawn care companies
Voice and chat intake that asks about the property and the turf, follow-up on quote requests, and a campaign for customers who stopped service.
See Lawn care companiesRestoration companies
Voice and chat assistants that take down a building restoration inquiry and answer the questions you set, with no price, no booking and no dispatch.
See Restoration companiesLead routing and call lists
Facebook leads land on one daily call list, each loan type goes to its own partner, and one AI agent speaks for whoever owns the lead.
See Lead routing and call listsEnglish and French AI agent
One voice agent for calls in both languages, plus AI chat, answering from one approved knowledge base. Built and tested, off until go-live.
See English and French AI agentPatient inquiries and clinical handoff
Chat and voice that take down each inquiry and a callback time, and hand every question about results or treatment to a person on your team.
See Patient inquiries and clinical handoffSales rep roleplay training
Reps pick a scenario and a difficulty, the agent answers as the prescriber, and every practice call ends with a recap.
See Sales rep roleplay trainingQuote request follow-up
Quote requests that arrive with the address and turf type, a welcome message that explains the estimate, and follow-up that mixes texts, emails and calls from staff.
See Quote request follow-upAI phone and chat intake
A voice assistant and a chat assistant that take down a restoration inquiry and answer your set questions, tried on a test page before your site.
See AI phone and chat intakeMissed-call AI receptionist
A Voice AI agent that picks up only when your team doesn't, with instructions that rule out score promises, deletions and fee talk.
See Missed-call AI receptionistGuides
Read first
AI agent API scopes and quirks
The exact scopes and parameters for HighLevel's Voice AI and Conversation AI endpoints, and the errors that come from mixing them up.
Read the guideWhat goes in an AI knowledge base
The entries worth writing, the source type that suits each, and the material that leads a bot to wrong or awkward answers.
Read the guideQuestions
What owners ask before an AI Employee setup
Which AI Employee plan does our HighLevel account need?
What does the AI say when it doesn't know an answer?
Can the AI answer in Spanish?
Can someone abuse an AI that answers questions all day?
Do we need chat and voice at the same time?
What happens to the AI's answers when we change a service?
Who you talk to
Rob Eason
Rob runs HLAI Experts and takes the 1:1 calls himself.
He is a HighLevel Certified Admin with the second-most reviews in HighLevel's Certified Admin Directory worldwide, all of them 5 out of 5.
He won HighLevel's Top New Subaccounts via AI Employee award in its 2025 Summer of AI Awards and taught HighLevel's official masterclass on its AI Employee.
Before HighLevel, he co-owned a 60-employee financial services firm and sold his share.
On the call he looks at your account, not a script.
5/5 across 83 reviews in HighLevel's Certified Admin DirectoryTalk through your AI Employee with Rob
Bring your website and the questions customers ask most. In 30 minutes Rob tells you what the AI should answer, what it should pass to your team, and which plan your account needs.