Joan AI: the office that runs itself, one sentence at a time

Visionect is submitting Joan AI — a conversational system that doesn’t just give a language model words, it gives it hands. It turns a chat message into a real action in the office: booking a desk, checking in a visitor, holding a meeting room, even starting the coffee machine.

Submitted by: Visionect (Joan Workplace)
Contact: Ana Križmančič, Marketing Director
For: AI Week 2026 program and awards

the PROBLEM

The office speaks ten different apps

Today

Five apps, ten minutes gone.

An employee opens one app to book a desk, another for the meeting room, a third to register a visitor, calls the front desk for coffee, and orders lunch through yet another platform. Every day, ten minutes disappear into admin that a single sentence could have handled.

With Joan AI

One conversation, real actions.

One conversation, in a tool the employee already uses (Teams, Slack, the web). Joan AI understands the intent and, through MCP (Model Context Protocol), actually carries out the booking, check-in, or order — in the real system, not just in a chatbot’s reply.

The idea, in one line: Language is cheap now. Hands in the physical world aren’t.

Anyone can string a sentence together today — Gemini, Claude, take your pick. What’s actually rare is a system that can do something on an employee’s behalf inside a real building: unlock a door, switch on the coffee machine, confirm a booking. Joan AI is that layer.

Joan AI isn’t an LLM. Joan AI is what gives an LLM hands.

Instead of Visionect hand-wiring every office device one at a time, we built an open connectivity layer — the MCP store. A company can plug in its own coffee machine, parking gate, or food delivery service; Joan discovers the new tool and simply offers it to every employee the next morning. This isn’t a smarter chatbot. It’s a system that physically reaches into the office and keeps growing without waiting on us.

The story: One Tuesday, through Maya’s eyes

Maya’s train pulls in four minutes late — she knows because she asked. Not the station app, not a browser tab — Joan, in the very same chat window where she talks to everyone else at work.

  • Check-in

    She checks in at the lobby with one line — no kiosk, no badge to fumble for. It’s the first handshake between digital Maya (calendar, team, preferences) and physical Maya, a person who needs somewhere to sit.

    “Let my company know I’m here”

  • The desk

    Joan puts her at the window desk it already knows she prefers, two seats from her colleague Luke, and drops the booking into her calendar without being asked twice.

    “Book me my usual spot, and is Luke in today?”

  • The coffee

    By the time she’s hung up her coat, the coffee is already brewing. Visionect never built a coffee-machine feature — someone in IT connected it themselves, the same afternoon it was installed. That’s the MCP store in practice.

    “Double espresso, I’ll grab it in two minutes.”

  • The team lunch

    At half past eleven, Joan checks who’s actually in the building, holds the table by the kitchen, and — because someone connected a food-delivery service months ago — orders lunch for eleven people to land at half twelve, warm. One sentence, zero apps opened.

    “Marketing and Product are both in today — find us a lunch space and sort out food.”

The philospohy: Employee is the center, not the system

We didn’t set out to build desk software, a room-booking tool, or a coffee integration. We set out to give employees back the minutes they lose to admin, and hand them to the work that actually matters. The features are just the means. Success isn’t measured by how many systems we connect — it’s measured by how little the employee has to think about them.

Why this fits AI Week 2026

A working, real-world example of applied AI

Icon

Build — Measurable business impact

Not a demo chatbot — a production system inside the Joan Workplace platform, already in use by real companies, with a clear time-and-admin saving for every employee.

Icon

Experience — A live, tangible demo

Joan AI is a natural fit for a live demonstration: a conversation that triggers a real, physical action — a booking, a check-in, an order — in front of the audience, in real time.

Icon

Awards — Slovenian-built, globally relevant

Built in Slovenia by Visionect, but architected on the open MCP standard and Gemini/Claude-class models — a case study in how a local company is solving a universal problem: the gap between the digital and physical workplace.

We’d love to show you Joan AI up close

We’d welcome the chance to present Joan AI as part of AI Week 2026 — as a business case (BUILD), a live demonstration (EXPERIENCE), or an entry for the awards. Happy to put together a deeper walkthrough or a custom demo for the organizers.

Ana Križmančič: Marketing Director, Visionect

Email: [email protected]

Company: Visionect · Joan Workplace