AI user journey generator
AI user journey generator for stages and satisfaction scores, written as a Mermaid journey diagram.
By MermaidViewer editorsPublished Updated
Example prompts for a user journey
- “Patient booking a doctor's appointment online”
- “First-time user setting up a smart home device”
- “Traveller checking in for a flight”
What a Mermaid user journey looks like
Example user journey
journey
title Signing up for a SaaS tool
section Discover
Find site via search: 4: Visitor
Read pricing page: 3: Visitor
section Sign up
Create account: 3: Visitor
Verify email: 2: Visitor
section First use
Complete onboarding: 4: User
Create first diagram: 5: User
Invite teammate: 3: User, TeammateWant to write it by hand? See the full Mermaid user journey syntax and examples, or browse templates.
A score of 5 on every step is a compliment, not research, and an ai user journey generator is only as honest as the numbers you insist on. Filing taxes, clearing an airport, returning a rental car: the diagram is stages, tasks, a score from 1 to 5, and the person who felt it. The widget above writes a Mermaid journey. This page covers prompts that keep the bad steps bad, and edits that change one score without a new generation.
Journeys the draft usually structures
Sections appear when you name stages: gather papers, file, wait for the refund. Tasks appear in that order. Actors stick when you repeat the actor on each task, which is what the syntax wants. A second actor, such as a gate agent or a rental clerk, shows up if you name them on the specific tasks where they join. Scores you state as digits are copied more reliably than scores you imply with words like "frustrating."
The chart is good at making a low number obvious. A 2 on "verify identity" is the slide that starts the meeting. The model will not discover that 2 for you. It will draw a 4 unless you said 2. The journey syntax is Task: score: Actor. Score is an integer from 1 to 5. There is no free-text sentiment field. If your research is quotes, you still have to pick a number before you generate.
Scores and actors it invents
Every task comes back as a 5, or as a smooth rise from 3 to 5, because the model wants the story to end well. That erases the reason you mapped the journey. Actors vanish on later tasks, so the legend shows a person who only existed in section one, and the tasks they still perform look unowned. A third actor gets added, "System" or "Team," and receives scores, which the syntax does not mean. The score belongs to the experience, and the names after it are who is involved.
Flowchart boxes appear when the prompt said "user journey" and the model heard "screens." You get decisions instead of feelings. Sections get skipped and twenty tasks sit in one row. Scores of 0, 6, or 3.5 fail or get rounded. Task names with a colon split the line into garbage. A missing title is legal and still a problem in a deck where three journeys sit next to each other.
The words that have to be in the prompt
A journey prompt is a scored script.
- A title that names the person and the episode.
- Sections, in order.
- Each task, the integer score from 1 to 5, and the actors on that task.
- An instruction to keep low scores low, repeating the digits.
- Actors who must not appear.
- A cap on tasks so the row stays readable.
The customer journey template follows a SaaS prospect from discovery through trial to purchase, with scores that mark friction. Generate a new map for taxes, an airport, or a car return. Open the template when the product is a signup funnel and you are replacing the sample scores with interview data. The prompt guide shows how edit differs from generate. Here, restating the low scores in the edit prompt is what stops the happy ending.
Three prompts
- User journey titled Filing a paper return. Section Prep: find last year's PDF, score 2, actor Taxpayer. Section File: copy numbers into the form, score 3, Taxpayer. Wait on hold for a PIN, score 1, Taxpayer. Section After: mail the envelope, score 3, Taxpayer. Track the refund, score 2, Taxpayer. Do not raise any score. Do not add an Accountant actor.
- User journey titled Getting through a midsize airport. Section Curb: drop bag with the airline, score 3, Traveler and Agent. Section Security: bin shuffle, score 2, Traveler. Secondary check, score 1, Traveler and Officer. Section Gate: buy water, score 4, Traveler. Board when the zone is called, score 3, Traveler and Agent. Keep the secondary check at 1.
- User journey titled Returning a rental car. Section Arrival: find the return lane, score 2, Driver. Section Counter: wait behind one couple, score 3, Driver and Clerk. Dispute a fuel charge, score 1, Driver and Clerk. Section Exit: walk to the shuttle, score 4, Driver. No other actors. Do not add a five-star final task called Satisfaction.
A low score left alone
The tax journey keeps the PIN call at 1. That is the step a service designer should see first. Nothing in the syntax celebrates it. The number does.
journey
title Filing a paper return
section Prep
Find last year PDF: 2: Taxpayer
section File
Copy numbers into the form: 3: Taxpayer
Wait on hold for a PIN: 1: Taxpayer
section After
Mail the envelope: 3: Taxpayer
Track the refund: 2: TaxpayerMove it to the editor when a second season of interviews changes "mail the envelope" from 3 to 4. Change the digit. Leave the task name if the task is the same.
Editing a journey that got cheerful
Use an AI edit when scores were flattened or a whole section was dropped. "Set Wait on hold for a PIN to 1 and Track the refund to 2. Remove the Accountant actor. Add section Prep before File. Keep every other score as written." Fixing a pattern across the file is what the generation is for.
Change the source when one digit is wrong. The line Wait on hold for a PIN: 1: Taxpayer is readable. Replace 5 with 1. Add an actor after a comma. Delete a task. Add a section line. Hand editing stays free. You should not spend a generation to record that the shuttle was fine.
A parse error on a journey line is usually a colon inside the task name, or a score that is not an integer. Shorten the task to a few words and put a digit between the colons. Do that in the editor. If the model returned a flowchart, an AI edit that says "rewrite as a journey with these exact scores" is the faster path back.
The allowance, stated for this chart
Five free AI uses, counted in total, not per day. The same five cover every diagram type. They are tied to the account, the IP address, and a device cookie. A journey needs a second use mostly when the first draft sets every score to 4 or 5. Take that use and paste the digits again. Then stop prompting.
Pricing is Starter at $6.99 a month with 500 AI conversations a month, or Pro at $11.99 a month with unlimited AI under fair use. Fair use is a ceiling for abuse, and hand editing never approaches it because hand editing stays free. After the five, you can still correct scores in the editor and download a PNG. Free is 1× with a watermark. JPG, SVG, and PDF are on Pro.
A flowchart will show the decision to dispute the fuel charge. It will not show that the dispute felt like a 1. If you need the feeling, stay with this generator. Other types are listed on the AI diagram generator page.
Frequently asked questions
How do I generate a user journey with AI?
Describe the user journey you need in the box above and click Generate. The AI writes valid Mermaid user journey code that renders instantly, and you can open it in the editor to refine it.
Is the AI user journey generator free?
You can try it free: 5 AI generations in total, not per day, with no sign-up. Starter includes 500 AI conversations a month and Pro is unlimited. Manual editing stays available, and every plan can still download a PNG. JPG, SVG and PDF are on Pro.
Can I edit the generated user journey?
Absolutely. The output is plain Mermaid code — edit it by hand in the live editor, or ask the AI for changes in natural language.