Back to Blog

How to Create an AI Employee: Step by Step

Henri Jung, Co-founder at Superkind
Henri Jung

Co-founder at Superkind

Superkind logo on an abstract orange waves texture

Creating an AI employee starts with a task you already do every week: answering the same customer enquiry, entering an order into two systems, or copying numbers into a Friday report. Monday morning, 08:30. Lisa, the operations lead at a growing agency, opens Outlook to 47 new messages. Half are quote requests she answers from the same template. The other half are order confirmations she copies into Asana and lexoffice by hand. By noon she has done two hours of work a person could explain to a colleague in five minutes. This guide shows the five steps to hand exactly that kind of work to an AI employee - with your approval wherever you want it, and without a single new tool to learn. Below you will see how Superkind’s ready-made AI team takes over inbox, orders, reports and leads automatically.

Key takeaways

Get a ready-made AI team - Superkind comes with AI employees (Inbox Manager, Order Tracker, Weekly Reporter, Lead Qualifier) that work in your programs automatically. Start for free with unlimited users, or build your own.

Keep control, keep data in Europe - hosted in Europe, GDPR compliant, no training on your data. You decide which steps need your approval before anything goes out.

Five steps, not a project - pick a task, name the AI employee, connect its channels, set approval rules, define a routine. The first result arrives in minutes, not months.

Work handed over, not a tool learned - you write to Superkind in Outlook, Teams or Slack like a colleague. It does the task in your programs, through 1,700+ integrations.

Your knowledge stays - every task makes the AI employee smarter. This knowledge (your Company Brain: customers, prices, processes) stays with the company, even when people leave.

Your AI team for everyday work

Re: Quote request facade cleaning
Katharina Berger, Hausverwaltung Nord
Could you send us a quote for the monthly facade cleaning of our office at Elbchaussee 12? Approx. 1,200 m2, starting January.
Inbox ManagerAI employee
New email in service@ mailbox
Reply drafted from your standard rates. Quote for 1,200 m2 facade cleaning attached as PDF, monthly net EUR 1,840. Filed in SharePoint under Hausverwaltung Nord.
QuoteEUR 1,840 net / month
Attachmentfacade-cleaning-quote.pdf
Filed inSharePoint > Quotes > 2026
Send quote to customerNew customer, so it waits for your OK.ApproveReject
Example: Inbox Manager at work in Outlook

What does it mean to create an AI employee?

An AI employee is a software agent that works inside your existing programs - email, Teams, Slack, your CRM, your accounting tool - like a colleague you can message. You assign it a task in plain language, it does the work in those systems, and you decide which steps need your approval before anything is sent, booked or changed. One example is Superkind: a ready-made AI employee that handles inbox, orders, reports and leads in Outlook, Teams and Slack, hosted in Europe.

The difference to a chatbot or a prompt in ChatGPT: an AI employee does not stop at answering a question. It enters the order in your project tool, drafts the quote from your rate card, files the attachment in the right folder and reminds you when a deadline is close. It works across systems, not inside one window.

How it works with Superkind: your AI team for everyday work

You write to Superkind by email, in Microsoft Teams, in Slack or in the Superkind web app, like you would write to a colleague, and Superkind does the task in your programs - through 1,700+ integrations including Pipedrive, lexoffice, Asana, Notion and Calendly. It works with your tools instead of replacing them, hosted in Europe and GDPR compliant.

With Superkind you don't start from zero: it comes with ready-made AI employees for everyday team work. They run automatically in the background, and you decide where they ask you first. Use them right away, adapt them, or build your own.

Inbox Manager

AI employee for email triage

Every morning 40 new messages, half of them requests that need a quick answer, the rest updates nobody reads. The Inbox Manager reads every incoming email, drafts replies for routine requests from your templates, files attachments in the right SharePoint folder and flags anything that needs you. Delivery confirmations and newsletter noise disappear on their own.

Works inOutlookSharePointMicrosoft Teams+ 1,700 more
Re: Quote request facade cleaning
Katharina Berger, Hausverwaltung Nord
Could you send us a quote for the monthly facade cleaning of our office at Elbchaussee 12? Approx. 1,200 m2, starting January.
Inbox ManagerAI employee
New email in service@ mailbox
Reply drafted from your standard rates. Quote for 1,200 m2 facade cleaning attached as PDF, monthly net EUR 1,840. Filed in SharePoint under Hausverwaltung Nord.
QuoteEUR 1,840 net / month
Attachmentfacade-cleaning-quote.pdf
Filed inSharePoint > Quotes > 2026
Send quote to customerNew customer, so it waits for your OK.ApproveReject
Example: Inbox Manager at work in Outlook

Order Tracker

AI employee for order entry

A signed contract comes in by email, and the order needs to land in your project tool and your invoicing system before the end of the day. The Order Tracker reads the signed document, creates the project in Asana, enters the order in lexoffice and posts a summary in the Slack channel so the team knows what is coming.

Works inGmailSlack+ 1,700 more
#orders
Order TrackerAI employee
New signed contract in inbox
New order from Hausverwaltung Nord entered. Project created in Asana, order logged in lexoffice. Kick-off scheduled for Monday.
ClientHausverwaltung Nord
ValueEUR 22,080 / year
Start1 January 2027
Runs automatically on every signed contract
Example: Order Tracker at work in Slack

Weekly Reporter

AI employee for reporting

Friday afternoon, the team lead needs the weekly numbers: open orders, hours logged, invoices sent, overdue payments. The Weekly Reporter pulls data from Asana, lexoffice and Pipedrive every Friday at 16:00, builds the report and posts it in Teams. No more copying between five tabs.

Works inMicrosoft TeamsExcel+ 1,700 more
Weekly ReportManagement
Weekly ReporterAI employee
Routine - every Friday at 16:00
Weekly report for 11-15 Nov ready. 3 new orders, 12 invoices sent, 2 overdue payments flagged.
New orders3 (EUR 58,400)
Invoices sent12 (EUR 41,200)
Overdue2 (EUR 3,800)
Runs automatically every Friday at 16:00
Example: Weekly Reporter at work in Microsoft Teams

Lead Qualifier

AI employee for lead qualification

New leads come in through the website form and sit in Pipedrive until somebody looks. The Lead Qualifier reads every new contact, checks the company size and industry against your ideal customer profile, scores it and routes hot leads to the right salesperson in Slack within minutes, not hours.

Works inSlackHubSpot+ 1,700 more
Superkind
New chat
My tasks
AI employees
Lead Qualifier
Inbox Manager
Order Tracker
Lead qualification rules

For every new lead in Pipedrive: check company size and industry, score against our ICP, and post hot leads in #sales on Slack. Leads under 10 employees go to the nurture sequence.

Reasoning ⌄
Lead Qualifier 09:00
New lead in Pipedrive

Rule saved. 4 new leads scored since yesterday. 1 hot lead posted in #sales: Müller Gebäudetechnik, 45 employees, facility management.

Hot leads1
Nurture2
Disqualified1

Runs automatically on every new lead

Send a message ...
Example: Lead Qualifier at work in Superkind

Every employee is ready to use right away. Or build your own on the platform.

Safe by default

Your tasks, customer data and company knowledge stay in Europe and are never used to train a model.

What you need

Checklist before the first AI employee

  • One task your team does every week that you could explain to a new colleague in five minutes (answering quote requests, entering orders, building the weekly report, qualifying leads)
  • The channel your team already works in: Outlook, Gmail, Microsoft Teams or Slack
  • Access to the programs where the task lives: your CRM (Pipedrive, HubSpot), your invoicing tool (lexoffice, sevdesk), your project tool (Asana, Notion, ClickUp) or your file storage (SharePoint, Google Drive)
  • A person who decides what needs approval and what runs on its own (the team lead, the owner, the department head)
  • A free Superkind account at app.superkind.ai - no credit card, no sales call, up to 150 hours of work free
  • Thanks to the 1,700+ integrations, these systems connect without custom code or new middleware

Step by step with Superkind

Step 1: Pick the first task

Choose one task your team repeats every week. Good first tasks: answering quote requests from the same template, entering confirmed orders into two systems, building the Friday status report, or sorting and routing new leads. Pick something with a clear trigger (a new email, a signed contract, a new lead) and a clear result (a quote sent, an order entered, a report posted).

  • Check - can you explain the task to a new colleague in under five minutes? Then it works as an AI employee task.
  • Tip - start with one task, not five. One working AI employee teaches you more than a plan for ten.

Step 2: Name the AI employee and assign its channel

Sign up at app.superkind.ai and give your first AI employee a name that says what it does: Inbox Manager, Order Tracker, Weekly Reporter or Lead Qualifier. Then connect the channel where the task starts: Outlook for email, Microsoft Teams for team messages, Slack for channel work or the Superkind web app.

TeamsLisa to Superkind

You are our Inbox Manager. Read every new email in service@nordlicht.de and draft replies for quote requests using our standard rate card.

Superkind now watches the mailbox and knows its role. One AI employee, one job - just like a colleague with a clear responsibility.

Step 3: Give it access to the right programs

Connect the programs where the task happens. For the Inbox Manager: Outlook for email, SharePoint for file storage, and Excel or lexoffice for the rate card. For the Order Tracker: Gmail, Asana and lexoffice. You pick which folders and data the AI employee may see and confirm the permissions - just like onboarding a new team member.

  • Check - the audit log records every file and system the AI employee accessed.
  • Tip - connect only the systems the task needs. Add more later as trust grows.

Step 4: Set approval rules

Decide which steps need your OK before the AI employee acts. For a quote request: new customers wait for approval, repeat customers run on their own. For orders: anything over EUR 10,000 waits for the team lead. For the weekly report: it goes out every Friday without asking, because it is internal.

SlackLisa to Superkind

Quotes for new customers: always wait for my approval. Repeat customers under EUR 5,000: send automatically.

The approval shows up wherever the approver works - in Teams, in Slack or in Outlook - with Approve, Reject or a comment. Everything else runs on its own.

Step 5: Test with a real task today

Give the AI employee a real task from your inbox right now. Forward a quote request, drop a signed contract in the channel or ask for the weekly numbers. Read the first result like you would read a draft from a new colleague: check the facts, the tone and the numbers. Every correction goes into the Company Brain and makes the next result better.

OutlookLisa to Superkind

Here is the quote request from Hausverwaltung Nord. Draft the reply from our rate card and wait for my OK before sending.

Answer: “Quote drafted. 1,200 m2 facade cleaning, monthly net EUR 1,840. PDF attached. Waiting for your approval.”

Set it up as a routine

Once the AI employee handles the task well, stop giving it one task at a time. Tell it once what should happen every day or every week, and it keeps running.

TeamsLisa to Superkind

Every morning at 08:00: read all new emails in service@, draft replies for quote requests, file attachments in SharePoint, and flag anything that needs me. Every Friday at 16:00: pull numbers from Asana, lexoffice and Pipedrive and post the weekly report in #management.

  • Daily - inbox sorted, routine replies drafted, attachments filed.
  • Weekly - the status report built and posted, without anyone copying between tabs.
  • On event - every new signed contract enters the order in Asana and lexoffice automatically.

When the work grows - more AI employees, more departments, a custom scoring model for leads - you do not start over. On the Superkind platform you build custom solutions and apps with their own interfaces, on top of everything Superkind already knows. The 1,700+ integrations keep everything connected to Pipedrive, lexoffice, Asana and SharePoint without new middleware.

Data protection and EU hosting

An AI employee sees customer data, prices and internal processes. That data belongs in Europe, and your team needs to know where it goes. Superkind is sovereign and independent: hosted and processed in Europe under European law, GDPR compliant, aligned with the EU AI Act (Regulation (EU) 2024/1689), with no training on your data. AI models from OpenAI, Anthropic, Mistral and Google are processed in Europe - no lock-in to one model vendor. Your company knowledge, routines and apps belong to you. A data processing agreement (DPA) is available through the Trust Center. For more on what sovereign AI means for your team, see our guide to sovereign AI.

Other ways to create an AI employee

WayEffortError riskBest for
SuperkindLow: message it in Outlook, Teams or SlackLow: approval where you want it, 1,700+ integrations, EU hostingTeams handing over real work across Pipedrive, lexoffice, Asana, Notion
ChatGPT / CopilotMedium: write prompts, copy results backMedium: no link to your systems, no approval stepOne-off text tasks, brainstorming
Zapier / MakeMedium to high: build flows step by stepLow in the flow, higher outside itFixed automations between two systems
Custom developmentVery high: months of dev workLow once built, high during developmentUnique processes no tool covers
Manual, no AIVery high: everything by handMedium: human errors, missed deadlinesTasks done once a year

For a broader look at AI tools built for Europe, see our overview. If you want to understand how your team’s knowledge feeds the AI employee, read the AI knowledge management guide.

Common mistakes

  • Starting with five tasks at once - one AI employee doing one task well teaches you more in a week than a plan for ten. Add the second task after the first one runs cleanly.
  • No approval rules from day one - without approval rules, the AI employee acts on its own for everything. Set the rules before the first real task: which steps wait, which run.
  • Giving it access to everything - connect only the folders and systems the task needs. Add more later as trust grows. The audit log shows what the AI employee accessed.
  • Never checking the first results - the first week is onboarding. Read every result like a draft from a new colleague. Every correction makes the Company Brain better.
  • Forgetting the routine - an AI employee that only works when you ask it is a chatbot with extra steps. Define when it runs on its own: every morning, every Friday, on every new lead.

6 tips for creating your first AI employee

1. Start with the task, not the technology

Pick the task that costs the most hours every week and that you could explain to a colleague in five minutes. That is your first AI employee.

2. Name it like a role, not a tool

“Inbox Manager” is clearer than “Email AI”. A name that says what the AI employee does helps the team trust it and know what to expect.

3. Put the approval where the approver works

If the team lead lives in Teams, approvals go to Teams. If the owner checks email first, approvals go to Outlook. Superkind uses the same rules in every channel.

4. Review the first week like onboarding

Read every result in the first week. Correct tone, numbers and details. Each correction feeds the Company Brain and improves every future task.

5. Make data protection part of your choice

Check where the AI processes your data, whether a DPA is available and whether your data is used for training. The EU AI Act (Article 4) requires AI literacy in your team - knowing where data goes is part of it3.

6. Plan for the second AI employee early

Once the first one runs, think about what else your team does by hand every week. The same setup - name, channel, programs, approval rules, routine - works for quotes, onboarding, support tickets, project reporting and more.

Conclusion

Creating an AI employee is not a technology project. It is picking one task your team does every week, giving it a name, connecting the right channels and programs, setting the approval rules and letting it run. The fastest way to see if it works: take the next quote request, order confirmation or report from your inbox and forward it to Superkind with one sentence. If the first draft matches what you would have written, you just saved the hours. If it does not, your correction makes the next one better - and the knowledge stays with the company.

Frequently Asked Questions

An AI employee is a software agent that works in your existing programs - Outlook, Teams, Slack, your CRM, your accounting tool - like a colleague. You assign it tasks, it does the work in those systems, and you decide which steps need your approval before anything goes out. Unlike a chatbot, it does not just answer questions: it enters orders, drafts quotes, files documents and builds reports.

Superkind connects through 1,700+ integrations to tools like Pipedrive, lexoffice, Asana, Notion, Shopify, Calendly, SharePoint and Excel. You pick the systems during setup and confirm which folders and data Superkind may access. Programs without an integration work natively via computer use - Superkind operates the browser or the desktop client like a colleague would.

Yes. You set approval rules per task type or per threshold. For example: quotes over EUR 5,000 wait for your OK, routine filing runs on its own. Superkind only sends, books or changes something after your approval on the steps you marked. Everything else runs automatically.

Superkind is hosted and processed in Europe, GDPR compliant and aligned with the EU AI Act (Regulation (EU) 2024/1689). It does not train on your data. AI models from OpenAI, Anthropic, Mistral and Google are processed in Europe. A data processing agreement (DPA) is available through the Trust Center at trust.superkind.ai.

You start for free with up to 150 hours of work, no credit card needed. The Business plan starts at EUR 50 per month for 50 hours of work a month, for the whole company with unlimited users - no price per seat. Unused hours carry over one month. Enterprise pricing is custom. As of November 2026.

Minutes. Sign up at app.superkind.ai, connect your first channel (Outlook, Teams or Slack), give Superkind its first task and watch it work. The first useful result typically arrives within the same day. Over the following days you add more systems, set approval rules and define routines.

Related Articles

Henri Jung, Co-founder at Superkind
Henri Jung

Co-founder of Superkind, the AI employee teams write to in Outlook, Teams and Slack. Henri helps companies, from small teams to the Mittelstand, hand their routine work to AI - without new tools, consultants or an in-house AI team.

Send Superkind your first task.

Start for free in Outlook, Teams or Slack. Or let us show you in 30 minutes how it runs in your systems.