Go from zero to a working University IT Support agent that answers from your help guides, creates Jira tickets automatically, hands off to other specialist agents, and finally runs as an automated workflow. No code. No setup.
⏱ 45 Minutes
🖥 Browser Only
🚫 No Code
🎯 Beginner Friendly
🎓 University IT Use Case
🧩 Agent · Orchestrator · Workflow
Two-Part Workshop
1
Mind Platform
Build a University IT Support agent — knowledge base, Jira actions, publish to the marketplace — then scale it up with an Orchestrator Agent and an AI Studio workflow.
2
Nouraa.AI
Build a voice-powered IT Support agent with a conversational flow — greeting, issue collection, Jira logging, and voice testing.
The Three Things You'll Build on Mind Platform
🤖
AI Agent
One specialist. It has a behavior prompt, a knowledge base it answers from, and actions it can take — like creating a Jira ticket. Steps 1–5.
🧩
Orchestrator Agent
A manager over several specialists. It reads the question, decides which agent should handle it, and routes the conversation there. Step 6.
⚙️
AI Studio Workflow
The same intelligence with nobody asking. An event fires — an email arrives — and the work happens on its own. Deploy one from a template. Step 7.
🧭
How to Choose Between Them
Use an agent when a person is asking questions. Use an orchestrator when you have more specialists than any single prompt should hold. Use a workflow when nobody should have to ask at all — the process should just run.
Before You Start — 60 Second Check
Part 1 — Agenda
Step
What You'll Do
Outcome
01
Login & navigate to Agent Marketplace
Explore available agents, tabs, and editor access
02
Create a University IT Support Agent — profile, prompt & starters
A working AI agent with IT help-desk behavior
03
Upload the University IT Support Knowledge Base PDF
Agent answers grounded in your real IT guides
04
Connect Jira — deploy "Create Issue" action
Agent can log IT problems as Jira tickets
05
Publish & test the agent end to end
Live agent that answers questions and creates tickets
06
Build an Orchestrator Agent over the IT + HR agents
One front door that routes each question to the right specialist
07
Deploy the Student Recruitment Agent workflow from a template
Incoming email answered with a drafted reply, automatically
Part 2 — Nouraa.AI Agenda
Step
What You'll Do
Outcome
01
Login to Nouraa.AI & navigate to Agents
Access the voice agent builder
02
Create a Conversational Flow agent & configure greeting
First node greets caller and asks about their IT issue
03
Add second conversation node to collect details
Agent asks for device, location, and symptoms
04
Add Extract Variable node
Issue description captured as a variable
05
Add Function node — IT Support ticket reporter
Agent logs the issue via a backend function
06
Add confirmation node & End Call
Agent confirms ticket and wraps up the call
07
Test with voice
Full voice agent working end to end
Part 1 · Step 1
Login & Navigate to Agent Marketplace
Open the Mind Platform, log in with your workshop credentials, sign into organization Mindplatform Sandbox, and explore the Agent Marketplace — the central hub where all agents live.
🏢
Sign into the workshop organization
After login, if you see an organization switcher, choose Mindplatform Sandbox. All workshop steps use this environment.
Follow each step
1
Open the platform in your browser
Go to the URL below and log in with the workshop credentials.
URLapp.mind-platform.ai
Emailadmincb@royalcyber.com
PasswordCyber@2028@!
2
Select the Sandbox Organization
If prompted with an organization switcher, select Mindplatform Sandbox to enter the workshop environment.
OrganizationMindplatform Sandbox
3
Open the Agent Marketplace
On the left-hand navigation pane, expand it and click Agent Marketplace. This is the central hub for all agents.
You should now see a grid of agent cards with a blue Create New Agent button in the top-right.
Don't see the nav pane? It may be collapsed to icons — click the ☰ / arrow at the top of the left edge to expand it.
4
Explore the Marketplace tabs
Inside the Agent Marketplace you'll see three tabs at the top:
Overview — Public agents available in your organization
Favourites — Agents you've pinned by clicking the heart icon
My Agents — Agents you can edit (cards show ⋯ three dots for edit options)
✅
You're In
You're now inside the platform and ready to create your first agent. Head to Step 2 to start building.
Part 1 · Step 2
Create Your University IT Support Agent
Build a University IT Support agent for students and staff. It will answer IT help questions from your knowledge base and log unresolved problems to Jira.
Follow each step
1
Click "Create New Agent"
In the top-right corner of the Agent Marketplace, click the blue Create New Agent button.
2
Choose agent type → AI Agent
You'll see two options: AI Agent and Orchestration Agent. Select AI Agent (we're building a single specialist agent). Click Continue in the bottom right.
Orchestration Agent is used when you want to connect multiple agents together — we won't need that today.
3
Enter the Agent Name
You're now on the Edit Agent → Profile screen. Enter the name below — replace "YourName" with your actual name.
Use your real name. Everyone in this workshop shares one sandbox. If you leave it as "YourName" you won't be able to find your own agent again in the marketplace — and you will need to find it in Step 6.
NameUniversity IT Support (YourName)
4
Enter the Description
DescriptionHelps students and staff with university IT questions and escalates problems by creating a Jira ticket when needed
5
Select Category → Operations
Click the Category dropdown and select Operations.
CategoryOperations
6
Paste the Behavior Prompt
Copy the full prompt below and paste it into the Behavior Prompt field. This defines how your agent thinks and responds.
Read it before you paste it. Notice the shape: it gives the agent a role, exactly two jobs, the single question it is allowed to ask, and a fixed confirmation format. Vague prompts produce vague agents — this specificity is what makes the agent behave the same way every time you demo it.
Behavior Prompt
You are a University IT Support Assistant.
You do two things:
1. ANSWER IT QUESTIONS
If someone asks about Wi‑Fi, email, passwords, VPN, LMS/Canvas, printing, device setup, or account access — answer directly from your knowledge base. Keep it short and practical. Include the next step they should try.
2. LOG AN IT ISSUE (JIRA)
If someone describes a problem you cannot fully resolve in chat (outage, broken laptop, access blocked, repeated errors):
- First respond with 2–3 lines: acknowledge the issue and give the single most helpful troubleshooting step right now.
- Then ask ONLY: "What is your name and student/staff ID so I can log this ticket?"
- Once they provide their name/ID, create the JIRA ticket immediately using their name/ID + their original message as the description. Do NOT ask any more questions.
- Confirm with:
"Issue logged ✅
Ticket: [MW-XX](https://royalcyber.atlassian.net/browse/MW-XX)
Our IT team has been notified."
IMPORTANT:
- Never ask them to repeat the full problem — they already described it in their first message
- For urgent security issues (compromised account, phishing), tell them to change their password and contact the IT help desk immediately
- Keep all responses short, friendly, and clear
7
Add Conversation Starters
These appear as clickable buttons when someone opens your agent. Add each one into the Conversation Starters field.
I cannot connect to campus Wi‑Fi in the library — it keeps disconnecting.
My university email password is not working and I cannot sign in.
Canvas will not load my course materials — I just see a blank page.
How do I set up VPN access from off campus?
8
Click "Create Agent"
Once all fields are filled in, click Create Agent. You'll land on the agent edit screen where you can continue configuring.
Your agent name appears at the top of the edit screen and a left panel shows Profile · Knowledge Base · Tools · Publish & Triggers. Keep this tab open — Steps 3, 4 and 5 all happen here.
Create Agent greyed out? A required field is still empty — Name, Description and Category are all mandatory. Scroll up and check each one.
💡
What You've Built So Far
A prompt-only IT Support agent. It can answer general IT questions, but it doesn't know your university's specific guides yet. Step 3 fixes that with a knowledge base.
Part 1 · Step 3
Add Knowledge Base — IT Support Guides
Download the University IT Support Knowledge Base PDF and upload it to your agent. From now on, every answer will be grounded in your actual IT guides and procedures — not generic AI knowledge.
Before (Step 2)
Agent answers from general AI training data. Helpful, but not grounded in your university's real IT policies and how-tos.
After (This Step)
Agent searches YOUR uploaded IT knowledge base in real time. Answers reference specific guides (e.g. Wi‑Fi setup, password reset, VPN, LMS access).
IT help desk hours, walk-up support locations, JIRA MW project for tickets, security@ and helpdesk@ contacts.
Follow each step
1
Download the IT Support Knowledge Base PDF above ↑
Click the ⬇ Download PDF button above to save Contoso_University_IT_Support_FAQ.pdf to your computer. You'll upload this file in the next step.
2
Navigate to Knowledge Base
In the agent edit screen, look at the left-hand panel. Below Profile (which you just filled), click Knowledge Base.
3
Click "Files" and upload the PDF
Inside the Knowledge Base section, you'll see data source options: Files, Web Pages, OneDrive, Google Drive, Box Drive, Microsoft Fabric. Click Files — a pop-up opens. Click Browse Files, select Contoso_University_IT_Support_FAQ.pdf, then click Upload File.
The file appears in the knowledge base list. Wait until its status shows as processed / ready — a document that is still indexing is not searchable yet.
Upload rejected or stuck? Re-download the PDF from the card above and try again — a partially downloaded file will fail to index.
4
Click Save, then Next
Click Save to apply changes, then click Next to proceed to the Tools section.
💡
This Is RAG — Retrieval-Augmented Generation
Your agent now searches the University IT Support knowledge base in real time and responds based on exactly what's written — referencing specific guides like campus Wi‑Fi and password reset. Your content. Your accuracy. Not generic internet knowledge.
Part 1 · Step 4
Action Agent — Connect Jira
Give your IT Support agent the ability to take real action. Connect Jira so it can automatically create tickets when a student or staff member reports a problem — not just answer in a chat window.
Knowledge Base Agent
Knows your IT guides. Answers accurately. But someone still has to manually create the support ticket in Jira.
Action Agent
Knows your guides AND takes action. User reports an IT problem → agent troubleshoots → creates a Jira ticket automatically when needed.
🔌
2,000+ App Connectors Available
Jira, Gmail, Outlook, Slack, Teams, Salesforce, ServiceNow, and more. Today we connect Jira — the same approach works for all of them.
Follow each step
1
Navigate to Tools
In the agent edit screen, click Tools in the left panel. You'll see three sections: LLM Settings (leave as default), Action Library, and API & MCP Server.
2
Expand Action Library → Search for Jira
Expand the Action Library section. You'll see 2,000+ available connectors. In the search bar, type Jira and click on it.
SearchJira
3
Connect your Jira account
If you already see admincb@royalcyber.com as a connected account, skip to Step 4. Otherwise:
1. Click Connect to Account
2. A pop-up says "Campusmind uses Pipedream to connect your account" → click Continue
3. Another pop-up opens → sign in with the Jira credentials below
4. Authorize the permissions (manage, view, update) → click Accept
Jira Emailadmincb@royalcyber.com
Jira PasswordCyber@2028@!
If asked for two-factor auth: Skip it — click through to continue.
4
Select the "Create Issue" action
From the list of available Jira actions, find and click Create Issue. You'll land on a configuration screen.
5
Configure the action parameters
Fill in each parameter by clicking Load Options on each field, then selecting from the dropdown:
Cloud IDRoyal Cyber
Project IDMind Platform Workshop
Issue TypeTask
Remember: Click Load Options on each field first — the dropdowns won't populate until you do.
6
Click "Deploy Action" → Next
Click the blue Deploy Action button. Your Jira "Create Issue" action is now live. Click Next to proceed to publishing.
The Jira Create Issue action shows as deployed in your agent's tool list. This is the moment your agent stopped being a chatbot and became an agent that acts.
Dropdown empty? You skipped Load Options on that field — click it, wait for the spinner, then pick the value. Deploy Action greyed out means one of the three parameters is still unset.
💡
Your Agent Can Now Act
It doesn't just answer IT questions — it can create a Jira support ticket automatically. A student reports Wi‑Fi down → agent troubleshoots, advises, and logs it. That's an Action Agent.
Part 1 · Step 5
Publish & Test Your Agent
Make your IT Support agent visible in the marketplace, then test it end to end — ask an IT question, get a guide-grounded answer, and create a real Jira ticket.
Follow each step
1
Enable Agent Marketplace
You should now be on the Publish & Triggers section. At the top, toggle on Agent Marketplace — this makes your agent visible to everyone in the organization.
2
Click Next → Finish
Click Next to proceed to User Access. No changes needed here — click Finish.
3
Chat with the agent
Click Chat with the agent in the top-right corner of the edit screen. Your agent is now live — try asking an IT question using one of the starters below.
I cannot connect to campus Wi‑Fi in the library — it keeps disconnecting.
My university email password is not working and I cannot sign in.
Canvas will not load my course materials — I just see a blank page.
How do I set up VPN access from off campus?
4
Ask the agent to create a Jira ticket
After the agent responds with guidance, ask it to create a ticket. The agent will log the issue in the Mind Platform Workshop Jira project.
Follow-upYes, please create a Jira ticket for this IT issue
5
Verify the ticket in Jira
The agent will return a link to the newly created ticket. Click the link — you'll land on the Jira project and can verify the summary, description, priority, and all the details are correct.
A real ticket now exists in the Mind Platform Workshop Jira project, created out of a conversation. Nobody filled in a form.
Agent answered but didn't create a ticket? It's following the prompt — it asks for your name and student/staff ID first. Give it those and the ticket is created immediately.
What you just did: Reported an IT problem → got guide-grounded help → created a real Jira ticket — all through a conversation with your AI agent. No forms. No manual data entry.
🎉
Part 1 Complete — You Built an AI Agent
In just a few minutes you went from zero to a working University IT Support agent that knows your guides, answers grounded questions, and creates Jira tickets automatically. No code required.
What You Built
01
AI Agent
Created a University IT Support agent with a behavior prompt and IT-focused conversation starters.
02
Knowledge Base
Uploaded the IT Support knowledge base — answers grounded in real Wi‑Fi, account, and LMS guides.
03
Action Agent
Connected Jira — agent troubleshoots IT issues and creates tickets in the Mind Platform Workshop project.
Part 1 · Step 6
Orchestrator Agent — One Front Door, Many Specialists
You now have an IT Support agent. The sandbox also has an HR Help Desk agent. Real people don’t know or care which one to open — they just have a question. An Orchestrator Agent sits in front of both, reads the question, and silently hands it to whichever specialist owns it.
Why Not Just One Big Agent?
One agent, one giant prompt
You bolt HR policy onto the IT prompt. Now the prompt is twice as long, the knowledge base mixes Wi‑Fi guides with leave policy, and answers start drifting between the two. Every new department makes it worse.
Orchestrator over specialists
Each specialist keeps its own tight prompt, its own documents, and its own actions. The orchestrator only decides who should answer. Adding a third department is adding one agent — not rewriting a prompt.
Input
Person asks
One question, in plain language. No menu, no department to pick.
→
Orchestrator
University Orchestrator
Reads intent against the handover descriptions and picks the owner.
→
Specialist A
University IT Support
IT knowledge base + Jira "Create Issue" action.
Specialist B
HR Help Desk
HR policy documents, warm and policy‑grounded.
⚠️
Prerequisite — do this first
An orchestrator routes to agents that already exist. Before you start, check both: your University IT Support (YourName) agent from Step 5 is created and published, and the HR Help Desk agent is visible in the Agent Marketplace. You will search for both by name in Step 6 below.
The Six Screens You’ll Move Through
Create New Agent→Orchestrator Agent→Profile→Connected Agents→Configure Handover→Publish
Follow each step
1
Click "Create New Agent"
From the home screen, go to the top right and click the blue Create New Agent button — the same button you used in Step 2.
2
Choose Orchestrator Agent → Continue
On the next screen you get the same two options as in Step 2. This time select Orchestrator Agent, not AI Agent. Then click Continue in the bottom right.
This is the option Step 2 told you to skip. An AI Agent does the work itself. An Orchestrator Agent decides which agent does the work.
3
Fill in the Profile — name, description, category
You’re now on the orchestrator’s Profile screen. Replace "YourName" with your own name so you can find it again in the shared sandbox.
NameUniversity Orchestrator (YourName)
DescriptionSingle front door for university questions — routes IT problems to IT Support and people questions to HR Help Desk
CategoryOperations
4
Paste the orchestrator prompt
This prompt does exactly one thing: it decides who answers. Notice what it does not contain — no Wi‑Fi advice, no leave policy. That knowledge lives in the specialists.
Orchestrator Behavior Prompt
You are the University Orchestrator — the single front door for every question a student or employee brings to the university.
You do not answer questions yourself. Your only job is to work out what the person actually needs and route them to the right specialist agent.
ROUTE TO "University IT Support" WHEN THE QUESTION IS ABOUT:
Wi‑Fi or the campus network, university email or Outlook, passwords, MFA, locked accounts, VPN or off‑campus access, LMS/Canvas, printing, laptops, device setup, or any software or system access problem.
ROUTE TO "HR Help Desk" WHEN THE QUESTION IS ABOUT:
leave (vacation, sick, maternity, paternity, bereavement), benefits (health, dental, vision, 401k, wellness perks), payroll (pay schedule, deductions, reimbursements), onboarding or offboarding, remote and flexible work policy, or code of conduct and workplace policy.
HOW TO ROUTE:
- If the request clearly belongs to one specialist, hand it over immediately. Do not narrate the routing or announce which agent you are using.
- If the request is genuinely ambiguous, ask ONE short clarifying question, then route.
- If a request contains both an IT part and an HR part, route the IT part first, then offer to help with the HR part.
- If it belongs to neither, say so in one line and point them to the main help desk.
NEVER:
- Never answer an IT or HR question from your own knowledge — the specialists have the documents, you do not.
- Never ask the person to repeat something they have already told you.
Why the routing lists are so specific. "Route IT questions to IT" is not a rule an agent can act on reliably — it has to guess what counts as IT, and it will guess differently each run. Listing the actual topics is what makes routing repeatable in front of an audience.
5
Add conversation starters — two IT, two HR
Still on the Profile screen. Deliberately mix the two domains: when an audience sees one agent answer both kinds of question, the point of orchestration lands without you having to explain it.
I cannot connect to campus Wi‑Fi in the library — it keeps disconnecting.
Canvas will not load my course materials — I just see a blank page.
How many vacation days do I have left, and do unused days carry over?
What is the policy for working remotely from another country?
Each pair is doing a different job. The first of each is a problem — on the IT side that can run all the way through to a Jira ticket. The second of each is a question, answered straight from the knowledge base. Four buttons, and the audience sees every capability you built.
6
Connected Agents — attach both specialists
Go to the Connected Agents tab in the tabs along the top. Search for each agent by name, tick both, then click Save.
Search 1University IT Support (YourName)
Search 2HR Help Desk
Both agents are ticked and saved. Your orchestrator can now reach either one.
Search comes up empty? Two things to check — the agent name must match exactly what you typed when you created it (if you left it as literal "YourName", rename it now), and the agent needs to be published: open it → Publish & Triggers → toggle on Agent Marketplace → Next → Finish.
7
Configure Handover — tell it when to use each agent
Go to the Configure Handover section. Each connected agent needs a handover name and description. Fill in both agents using the copy below.
🎯
This Is the Step That Decides Routing Quality
The handover description is what the orchestrator actually reads when choosing where to send a question — often weighing it more heavily than the prompt itself. "IT stuff" and "HR stuff" will route the easy questions correctly and coin‑flip the interesting ones. Paste these in full.
Handover 1 — University IT Support
Handover NameTransfer to University IT Support
Handover Description — IT
Hand over to this agent for any technology, device, or system access problem. This includes campus Wi-Fi and network connectivity, university email and Outlook, passwords, MFA, locked or blocked accounts, VPN and off-campus access, LMS and Canvas, printing and print queues, laptops and device setup, software installation, and any error message coming from a university system. This agent answers from the official IT support guides and can log the problem as a Jira support ticket when it cannot be resolved in the conversation. Do not use this agent for questions about pay, leave, benefits, or workplace policy.
Handover 2 — HR Help Desk
Handover NameTransfer to HR Help Desk
Handover Description — HR
Hand over to this agent for any people, policy, or employment question. This includes leave of every kind (vacation, sick, maternity, paternity, bereavement), benefits (health insurance, dental, vision, 401k, wellness perks), payroll (pay schedule, deductions, reimbursements), onboarding and offboarding procedures, remote and flexible working policy, and code of conduct or workplace policy. This agent answers from the official HR policy documents and refers the employee to hr@mind-platform.ai when something is not covered. Do not use this agent for technical problems with devices, accounts, or university systems.
Notice the last line of each description. "Do not use this agent for…" is doing real work. Without it, "I can’t log into the payroll portal" is a coin flip — it has payroll words in it, but it is an account access problem. With the boundaries in place it lands on IT, which is correct.
Name field rejecting spaces? Use transfer_to_it_support and transfer_to_hr_help_desk instead — the description is what matters for routing, not the name.
8
Publish — toggle Agent Marketplace, then Save
Go to the Publish section, toggle on Agent Marketplace, and click Save.
Your University Orchestrator is live in the marketplace, alongside the two specialists it manages.
9
Save your agent
Click Save. That is the build finished — profile, prompt, starters, connected agents, handovers and publishing are all in place.
Your orchestrator is saved and live in the Agent Marketplace.
10
Find your agent in the marketplace
Go to the Agent Marketplace and search for your orchestrator by name. Open it, and start a conversation.
SearchUniversity Orchestrator (YourName)
Not showing up? The Agent Marketplace toggle in the Publish section didn’t save. Reopen the agent, toggle it on, and click Save again.
11
Test the routing — the moment that sells it
Click one of the conversation starters you added in Step 5, or type a question of your own. Never tell it which agent to use — that is the entire point. Watch it hand off to the right specialist and bring the answer back into the same conversation.
Say this
Should route to
My VPN stopped working from home.
IT Support
What is the parental leave policy?
HR Help Desk
Canvas will not load my course materials.
IT Support
When does my health insurance start?
HR Help Desk
New joiner — laptop setup and benefits enrolment.
IT first, then HR
Every question landed with the right specialist and came back into one conversation — and the IT ones can still create a Jira ticket. The orchestrator cost you none of the capability you built in Steps 3 and 4.
Everything routing to one agent? Two usual causes: only one agent actually saved in Connected Agents (go back to Step 6 and confirm both are ticked), or a handover description got truncated on paste — reopen Configure Handover and check both are complete, including the closing "Do not use this agent for…" line.
🎉
You Built a Multi‑Agent System
One conversation, two specialists, correct handoffs, and real actions underneath — and nothing you built in Steps 1–5 had to change to get here.
The ceiling is a lot higher than two. Finance, Facilities, Admissions, Library, Student Records are each one more agent attached to the same front door — never a rewrite of the ones already working. That is how a genuinely complicated, powerful scenario gets built on this platform: not one enormous agent, but many small ones that each do a single job well, behind one door.
Reference — HR Help Desk Prompt
You only need this if the HR Help Desk agent is missing from your sandbox. Create it exactly like Step 2 — AI Agent → name it HR Help Desk → category Operations → paste the prompt below → upload your HR policy document as its knowledge base.
HR Help Desk — Behavior Prompt
You are a knowledgeable and empathetic HR assistant for Mind Platform.
You help employees with questions related to:
- Leave policies (vacation, sick leave, maternity, paternity, bereavement)
- Benefits (health insurance, dental, vision, 401k, wellness perks)
- Payroll (pay schedule, deductions, reimbursements)
- Onboarding and offboarding procedures
- Remote work and flexible work policies
- Code of conduct and workplace policies
When responding:
1. Always search the HR policy document first before answering
2. Be warm, professional, and empathetic in your tone
3. Reference the relevant policy section in your answer
4. If a question is outside of HR, politely redirect the employee to the right department
Never speculate or make up policy details. If the answer is not in your documents, let the employee know and advise them to contact HR directly at hr@mind-platform.ai.
Part 1 · Step 7
AI Studio Workflow — An Agent That Works Your Inbox
Everything so far needed somebody to open a chat and ask. This doesn’t. A prospective student emails the university, and a drafted reply is already waiting in Gmail before anyone has read it. Same intelligence — no conversation, no one asking.
Agent vs Workflow — The Actual Difference
Agent — reactive
Waits to be asked. Somebody has to open it, type a question, and read the answer. Perfect when a human is already in the conversation.
Workflow — triggered
Fires on an event, not a question. An email lands and the work happens. Perfect when the process is known and nobody should have to ask for it.
🎯
What We’re Automating
Admissions inboxes get the same questions every day — entry requirements, deadlines, portal logins — and every one waits for a human to sit down and type. This workflow reads each incoming email and leaves a drafted reply in Gmail. A person still reviews and sends. The blank page is what got automated, not the judgement.
⚡
Good News — You Are Not Building This One From Scratch
AI Studio ships with templates. You’ll take the Student Recruitment Agent, connect your Gmail, and publish it. Most of this step is authorising an account — the workflow itself already exists.
What You’re Deploying
Trigger
Email arrives
Gmail watches the inbox. Nobody opens anything.
→
Read
Gmail — get message
Pulls the email body through as plain text.
→
Think
Student Recruitment Agent
Works out what was asked and composes the reply.
→
Act
Gmail — create draft
Leaves the reply as a draft for a human to send.
Follow each step
1
Open AI Studio → Browse Templates
In the left-hand navigation pane, click AI Studio. Then click Browse Templates. You’ll land on a wall of template cards — prebuilt workflows you can deploy rather than design.
You’re looking at template cards, not the agent editor. Different surface, same platform.
2
Find the Student Recruitment Agent → Use Template
Search for the template below, or scroll until you find its card. Click Use Template on that card.
SearchStudent Recruitment Agent
A pop-up opens and walks you through connecting the accounts this workflow needs. The next few steps are all inside that pop-up.
3
Connect your Gmail account
The pop-up asks you to sign in to Gmail. If no account is connected yet:
Click Connect to Gmail
Click Connect an Account
An OAuth window opens — sign in and Allow the permissions
Your Gmail account shows as connected in the pop-up.
Nothing happened when you clicked? The OAuth window is a browser pop-up — if pop-ups are blocked it is silently suppressed. Allow pop-ups for this site and click again.
4
Continue → Continue to configure the action
With the account connected, click Continue. Then click Continue again to reach the action configuration screen.
5
Enable "Return payload as plain text"
This is the only setting you need to change in the whole workflow. On the Gmail configuration screen, find the field below and switch it on. Everything else is already correct.
EnableReturn payload as plain text
Why this one toggle matters. Left off, the agent receives the raw email payload — headers, encoding, MIME parts — instead of the message a person actually wrote. It will still produce a draft, just a confused one. This is the single most likely thing to go wrong today, and it fails quietly.
The pop-up confirms Gmail is ready. Click Done.
6
Configure the second Gmail action — Create Draft
The pop-up moves on to the Create Draft action. Your account is already connected from Step 3, so there is nothing to authorise — click Continue, then Continue again.
Leave every field empty. They are all optional, and the agent fills them itself — recipient, subject and body all come from the email it just read.
This is worth pausing on in a demo. You are not mapping fields or writing templates. The agent decides who to reply to and what to say, because it has already read the message. Filling these in by hand would make it less capable, not more.
The pop-up confirms Gmail is ready — click Done. You should now see All setup complete.
7
Continue — and watch it deploy
Click Continue. The platform now builds and ships the workflow for you. It takes about 30 seconds, and it shows you each stage as it goes:
Configuring Action→Publishing Workflow→Deployment Triggers→Workflow is Live
The workflow is published and live. It is now watching that inbox — nobody has to open anything for it to run.
Don’t skip past the progress bar in a live demo. Those four stages are the difference between a chatbot and a deployed system. Let the room watch it happen.
8
Test it — send yourself an email
Send an email to the connected account asking about admissions or IT. Use one of these, or write your own.
Test Email — Admissions
Subject: Question about the MSc Data Science programme
Hello,
I am interested in applying for the MSc Data Science programme starting next September. Could you tell me what the entry requirements are, whether you accept international applicants, and when the application deadline is?
Thank you,
Priya
Test Email — IT / Portal Access
Subject: Cannot access the applicant portal
Hello,
I created an account on the applicant portal last week, but now I cannot log in — it says my account is locked. Could someone help me reset it so I can finish my application?
Thanks,
Daniel
Give it a moment. The trigger polls the inbox rather than firing the instant you hit send. If nothing appears immediately, wait and check again before assuming it failed.
9
Check the draft — and the run notes
Open Drafts in Gmail. There is a reply waiting that you did not write, addressed to the person who emailed and answering what they actually asked. Read it, adjust anything you want, and send.
An email arrived and a considered reply was ready before a human opened the inbox. That is the whole point of a workflow in one screenshot.
Back in AI Studio, open the workflow’s run history to see the notes for that execution — what the trigger picked up, what the agent decided, and what it wrote. Useful for debugging, and useful for showing a sceptical room that nothing magic happened.
Draft is generic or garbled? Almost always the Return payload as plain text toggle from Step 5 — the agent got raw payload instead of the message. Reopen the Gmail action, switch it on, and republish.
🎉
Part 1 Complete — Agent, Orchestrator, Workflow
You built all three shapes Mind Platform offers, in one sitting, without writing code — and the last one you deployed from a template in about the time it takes to make coffee.
What You Built Today
01
AI Agent
A University IT Support agent with a knowledge base and a live Jira action — it answers, and it acts.
02
Orchestrator
A University Orchestrator routing between IT Support and HR Help Desk, with no menu for the user to navigate.
03
Workflow
A Student Recruitment workflow watching an inbox and drafting replies to incoming email, with nobody asking.
Now Build Your Own — Hands‑On Lab
Pick a process from your own team and rebuild the pattern. You have everything you need — don’t start from a blank page, start by copying what you just made and changing the domain.
Part 2 · Step 1
Login & Navigate to Agents
Open Nouraa.AI — a voice-first AI agent builder — and navigate to the Agents section where you'll build a conversational University IT Support voice agent.
🎙️
What Is Nouraa.AI?
While Mind Platform builds text-based AI agents, Nouraa.AI lets you build voice-powered AI agents using a visual conversational flow builder. Same University IT Support use case — but now accessible via a phone call.
Follow each step
1
Open Nouraa.AI in your browser
Go to the URL below.
URLapp.nouraa.ai
2
Login with your email
Enter the email below, then enter the verification code when prompted.
Emailadmincb@royalcyber.com
Code234567
3
Navigate to Agents
Once you're on the home screen, expand the left-hand navigation pane and click Agents. This is where you'll create your voice agent.
✅
You're In
You're now inside Nouraa.AI and ready to build a voice-powered conversational agent. Head to Step 2 to start.
Part 2 · Step 2
Create Flow Agent & First Node
Create a new Conversational Flow agent, name it, and configure the first node — the greeting that callers hear when they reach University IT Support.
Follow each step
1
Click "Add New Agent"
In the top-right corner, click the blue Add New Agent button. A pop-up opens with two options.
2
Choose "Conversational Flow" → Create Agent
Select Conversational Flow (not Single Prompt Agent) and click Create Agent. A simple flow appears with a Start node and a Conversation node.
Conversational Flow lets you design a multi-step voice conversation visually — each node handles one part of the dialogue.
3
Rename your agent
At the top of the screen, you'll see the name New Flow Agent with an edit pencil icon next to it. Click the pencil, enter the name below, and click the checkmark to confirm.
Agent NameUniversity IT Support (YourName)
4
Open the first Conversation Node settings
Click the three dots on the existing Conversation node to open its settings.
5
Enter the greeting message
In the Message field, paste the greeting below. This is what callers hear first.
Greeting Message
Hello, you've reached University IT Support.
Are you calling about an IT problem or need help with something on campus?
6
Keep default settings
Leave Skip Response and Global Node as default (both turned off).
7
Add a Condition → Save
In the Conditions section, click Add+ and enter the condition below. Then click Save.
ConditionNeed help with an IT problem
💡
First Node Done
Your agent now greets callers and waits for them to describe their IT issue. Next, we'll add a node to collect the details.
Part 2 · Step 3
Collect IT Issue Details
Add a second Conversation node that asks the caller to describe their problem — including device, location, and what they're seeing.
Follow each step
1
Add a new Conversation Node
On the bottom pane, click the + icon. You'll see multiple node types: Conversation, Function, Extract Variable, etc. Click Conversation Node. A new node appears on the canvas.
2
Connect the nodes
The first Conversation node has a blue dot at the bottom. The new node has a blue dot at the top. Click and drag between the two blue dots to connect them.
3
Open settings & enter the message
Click the three dots on the new node. In the Message field, paste the text below.
Detail Collection Message
Sorry to hear that. Please describe your IT issue — what device you're using, where you are on campus, and any error messages you see.
4
Keep defaults & add Condition → Save
Leave Skip Response and Global Node off. In Conditions, click Add+ and enter:
ConditionUser describes IT issue details
Click Save.
✅
Two Nodes Connected
Your agent now greets the caller, then asks for IT issue details. Next we'll extract the description into a variable so the function can use it.
Part 2 · Step 4
Extract Variable — Issue Description
Add an Extract Variable node to capture the caller's IT issue description as a structured variable. This feeds into the function that logs it to Jira.
Follow each step
1
Add an Extract Variable node
Click the + icon on the bottom pane → select Extract Variable. A new node appears on the canvas.
2
Open settings & add a variable
Click the three dots on the node. In the Variable section, click + to add a new variable. A pop-up opens.
3
Configure the variable
Fill in the fields below. Variable Type should be Text.
Variable Nameincident_description
DescriptionDescription of the university IT support issue
TypeText
4
Save variable → Save node
Click Save on the variable pop-up, then click Save again on the whole node.
5
Connect the node
Connect the blue dot at the top of this node to the blue dot at the bottom of the previous Conversation node.
💡
Variable Captured
The caller's issue description is now stored as incident_description. This variable is passed to the function in the next step.
Part 2 · Step 5
Function Node — IT Support Ticket Reporter
Add a Function node that takes the captured issue description and logs it as a support ticket. In Nouraa, search for HSE Incident Reporter (workshop backend name) or your IT ticket function.
Follow each step
1
Add a Function node
Click the + icon on the bottom pane → select Function. A pop-up opens asking you to select a function.
2
Search and select the IT Incident function
In the search bar, type IT Incident. Find and click IT Incident. The node appears on your canvas.
SearchIT Incident
3
Enable Ambient Sound
Click the three dots on the node to open settings. Under Settings, find Ambient Sound and enable the toggle. You'll see Keyboard Typing appear as the ambient sound — this plays while the function runs so the caller hears activity.
Why ambient sound? The function takes a moment to create the Jira ticket. The typing sound lets the caller know something is happening rather than hearing silence.
4
Save the node
Click Save.
5
Connect via the Success path
The previous Extract Variable node has two output dots: Success and Fail. Connect the Success path to the blue dot at the top of this Function node.
✅
Backend Connected
When a caller describes an IT issue, the function automatically logs it as a Jira ticket. Next, we'll confirm the ticket and end the call.
Part 2 · Step 6
Confirmation & End Call
Add a final Conversation node to confirm the ticket was created, ask if the caller needs anything else, and then an End Call node to wrap up.
Follow each step
1
Add a Conversation Node
Click + on the bottom pane → select Conversation Node. Connect it to the Success path of the Function node.
2
Enter the confirmation message
Click the three dots to open settings. In the Message field, paste the text below.
Confirmation Message
Logging your issue was successful. Here is your ticket number.
Is there anything else I can help you with?
3
Add Condition → Save
In Conditions, click Add+ and enter the condition below. Then click Save.
ConditionUser doesn't need anything else
4
Add an End Call node
Click + on the bottom pane → select End Call. The End Call node appears on the canvas.
5
Connect to the End Call node
Connect the blue dot from the Confirmation conversation node to the red dot of the End Call node.
Your entire voice agent workflow is now configured: Greeting → Collect Details → Extract Variable → Function (Jira) → Confirm → End Call
🎉
Flow Complete
Your voice agent workflow is fully built. All nodes are connected. Head to Step 7 to test it with your voice.
Part 2 · Step 7
Test Your Voice Agent
Run your completed flow using voice testing. Speak to the agent, describe an IT problem, and watch it flow through every node — from greeting to Jira ticket creation.
Follow each step
1
Click the Play button
On the top-right toolbar, click the Play button. A testing pop-up opens.
2
Switch to Voice Testing → Start
In the testing pop-up, you can test by typing or by voice. Click Voice Testing, then click Start.
3
Have a conversation with your agent
The agent will walk through the entire flow:
Node 1Agent greets you and asks about your IT problem
Node 2Agent asks for details — device, location, and error messages
Node 3Extracts your issue description into a variable
Node 4Runs the ticket reporter function (keyboard typing sound plays)
Node 5Confirms the ticket was created and asks if you need anything else
EndCall ends
🎉
Workshop Complete!
You've now built two AI agents — a text-based University IT Support agent on Mind Platform that answers from IT guides and creates Jira tickets, and a voice-powered IT Support agent on Nouraa.AI that handles the same workflow through a phone conversation. No code. No engineering required.
What You Built Today
🖥️ Part 1 — Mind Platform
A text-based University IT Support agent with a knowledge base (IT help guides) and a Jira "Create Issue" action. Students and staff chat with the agent, get guide-grounded answers, and log IT problems as Jira tickets.
🎙️ Part 2 — Nouraa.AI
A voice-powered IT Support agent built with a conversational flow. Callers describe an IT problem by voice, and the agent automatically logs it to Jira — with ambient typing sounds while it works.
Same Use Case. Two Interfaces. Zero Code.
Whether users prefer chat or voice, the AI agent handles IT support questions, troubleshooting guidance, and Jira ticket creation — automatically, 24/7.
