Customer Journey AI turns your enrollment pipeline into an automated, audited, AI-assisted workflow — from the moment a prospect submits a lead form to the day they enroll. Your team configures the journey once: the stages, the rules for moving between them, and the AI agents that handle outreach and questions along the way. The AI does the routine work 24/7. Your staff stays in control of everything that matters.
This article explains how Customer Journey AI works, what AI Teams are and how they fit in, and how to get started.
The mental model
Customer Journey AI has two layers that work together:
Customer Journeys are the map — the stages a contact moves through (Inquiry → Application → Enrolled), the rules that move them, and the outreach that happens at each step.
AI Teams are the staff — the AI agents that do the talking, answer questions, and make judgment calls across SMS, email, voice, and chat.
A Journey without AI Teams is just an automated pipeline. AI Teams without a Journey are just chatbots. Together they give your school a 24/7, on-brand student engagement workforce that scales without adding headcount.
What it solves
Most admissions teams are small — one to three people managing hundreds of leads. Leads go cold fast, follow-ups slip, and the tools that exist today are deterministic: send an email when a form is submitted, then stop. They can't decide who to contact, what to say, or whether now is the right moment.
Customer Journey AI does that. It keeps every prospect moving through your pipeline at the right pace, through the right channels, with a consistent voice — and logs every decision so you can see exactly why it did what it did.
How a journey works
A journey is built from four building blocks:
Stages — the phases in your enrollment lifecycle. Examples: New Inquiry, Application Started, Tuition Deposit, Enrolled. Each stage can have a student-facing label (what students see in their progress portal) and can include a self-serve action the student completes themselves.
Plays — the outreach attached to each stage. There are three types:
Manual — surfaced to a staff member to handle personally
Auto — a deterministic message sent automatically (SMS, email, task, or call) based on a template
AI — an AI agent executes the outreach using your instructions and your school's knowledge base
Transitions — the rules that move a contact from one stage to the next. Three types:
Manual — a staff member moves the contact in the pipeline
Auto — a rules engine evaluates conditions (e.g., "3 days in stage with no reply") and moves them automatically, checked every minute
AI — an agent reads the decision criteria you set in plain language, reviews the contact's context, and decides whether to advance them
Stage Actions — self-serve tasks sent to students via a magic link: pay an invoice, upload a document, or book an appointment. When a student completes the action, the journey advances automatically.
How a contact moves through a journey
Here's the full picture end-to-end:
A prospect submits a lead form → they're automatically placed into your journey at the first stage
An AI Play fires — the Enrollment agent sends a personalized SMS grounded in your school's knowledge base
The prospect replies with a tuition question → the Customer Service agent answers in real time from your knowledge base
An AI Transition evaluates overnight whether the prospect qualifies to move to the next stage
At the deposit stage, a magic-link pay invoice is sent → payment triggers an automatic advance to Enrolled
Every step — every AI decision, with its reasoning — is written to the Activity Log
Your team can watch the pipeline in real time, intervene with a manual move at any time, and see exactly why the AI made every call it made.
AI Teams — the agents behind the journeys
AI Teams are three groups of AI agents, each with a different role:
Team | What they do |
Marketing | Outbound outreach to prospects — re-engagement, newsletters, event invites |
Enrollment | Moves students from inquiry to enrolled — follow-ups, qualification, scheduling |
Customer Service | Answers inbound questions 24/7 — routes, transfers, and keeps students informed |
Each agent has a name, a persona, a tone, and a Knowledge Base — a searchable set of sources (your website, uploaded documents, Q&A pairs) the agent uses to answer questions accurately. You control what the agent knows and how it behaves.
Agents can work across SMS, email, voice, and chat. Each agent has a distinct voice profile for phone calls.
Agents disclose themselves as AI — there's a per-agent setting that controls this, and it's on by default. Don't turn it off without understanding your local disclosure requirements.
What the AI can and can't do autonomously
Not every action is treated equally. You control the balance:
Type | What it covers | What happens |
Autonomous | Routine outreach — a follow-up SMS, answering a standard question | AI executes and logs it |
Confirm | Higher-stakes actions — a campaign blast, contacting a specific person | AI proposes; a staff member confirms before it goes |
Task created | Things that need a human — a tour request, a special circumstance | AI creates a task with full context for your team |
Hard block | Compliance lines — contacting an opted-out contact, calling a DNC number | AI cannot execute; no override |
Manual plays and transitions always exist alongside the AI — you're never locked out of your own pipeline.
Important things to know about timing
AI transitions evaluate nightly (around 10 PM), not in real time. If a school expects a contact to move stages the moment something happens, that expectation needs to be set correctly. The exception: inbound replies trigger an AI check immediately.
Auto transitions run every minute — rules-based movement is near real-time.
Knowledge base changes aren't instant — when you add a new website or document source, it takes time to crawl and process before the agent can answer questions about it.
Draft vs. Live — changes to a Journey or an Agent don't go live until you publish them. If a school says "I made a change and nothing happened," check whether they published it.
Reviewing what the AI is doing
Under Customer Journey in your dashboard you'll find:
Journeys — list, create, duplicate, and publish your journeys; browse platform templates
Activity Log — every action and AI decision across every journey, with the AI's reasoning
Transition Evaluations — a full record of every AI transition decision and why
Dashboard — conversion funnel and activity metrics across your pipeline
Teams / AI Knowledge — manage your agents, their personas, and their knowledge bases
You can pause, edit, or cancel any journey at any time.
Testing before you go live
Use the Simulators before activating anything with real contacts:
Journey Simulator — walks through a journey with a test contact so you can see what fires when
Transition Simulator — tests your decision criteria with a specific contact and shows the AI's reasoning
Play Simulator — previews what a play will do before it runs
Try-Agent sandbox — have a real conversation with your agent before it talks to any students
Coach schools to use these heavily. It's the fastest way to build trust with the system.
Setting it up
Go to School Dashboard → Customer Journey
Create a new journey or start from a platform template
Build your stages — give each one an internal name and a student-facing label
Add plays to each stage (Manual, Auto, or AI)
Configure transitions between stages — set your rules or AI decision criteria in plain language
Set up your AI Teams: go to Teams, customize each agent's persona, tone, and guardrails
Add sources to your Knowledge Base — your website, uploaded documents, or Q&A pairs
Use the Simulators to test the journey end to end
Publish your journey and activate
AI Teams and other Lumion products
AI Teams currently powers Customer Journey AI — that's the primary product using the agent layer today. Additional Lumion products will build on AI Teams in the future, so it's worth understanding agents and knowledge bases as a foundation, not just as a CJAI feature.
What's not available yet
SIS and external system integrations
State-specific AI disclosure tooling (schools need to comply manually where required)
Stage branching and fork paths (e.g., separate Title IV vs. cash-pay tracks)
Financial aid, housing, and job placement agent templates
Cross-school benchmarking
Full IVR-style inbound decision trees
These are planned for later phases.
FAQ
Will the AI message students without my approval?
Routine outreach (follow-up texts, standard question replies) runs autonomously and is logged. Higher-stakes actions require your confirmation first. Hard compliance lines are always blocked. Manual plays and transitions exist alongside AI so your team is never locked out.
Can multiple journeys run at the same time?
Yes. Multiple journeys can run across different contacts simultaneously.
Can a contact be in two stages of the same journey at once?
No — a contact has exactly one active position per journey. They move forward one stage at a time.
What if the AI says something wrong?
Every action is logged with the AI's reasoning. You can flag it, update the agent's knowledge base or instructions, and republish. The Activity Log is your audit trail for exactly this.
Do students see the stage names and AI reasoning?
No. Students see a progress portal with the student-facing labels you set, and receive magic-link actions. They don't see stage names, plays, or AI decision reasoning.
What happens when a magic link expires?
Magic links expire after approximately 14 days by default. If a student tries to use an expired link, they'll be prompted to contact the school. You can trigger a new one manually.
What if a contact goes quiet for a long time?
Stages can have an off-ramp configured — if a contact is idle for a set number of days, they're automatically moved to another journey (or exited). This is intentional, not a bug — make sure you've configured off-ramps deliberately.
Can I disable AI for a specific contact?
Yes. Mark the contact as "Human handling only" on their profile. The AI will create tasks for your team instead of acting directly.
Does voice work on every agent?
Voice is available and each agent can have a distinct voice profile. Confirm with your Lumion rep that voice is enabled for your account.
How is this priced?
Check with your Lumion account manager for current pricing.
Troubleshoot
If the AI does something unexpected:
Open the Activity Log and find the action
Review the AI's reasoning — it's shown inline
If it's wrong, flag it and update the agent's knowledge base or instructions
Related articles
AI Teams: Getting Started (see note below)
AI Scheduler: Overview
SMS Marketing: Getting Started
Lead Capture AI: Getting Started
Lumion Inbox: Getting Started
Note: AI Teams and Customer Journey AI are deeply connected — AI Teams is the agent layer that powers the AI parts of a Journey. A dedicated AI Teams article is planned. For now, refer to this article for agent setup and knowledge base configuration.