Grega Pusnik
“AI makes it easier than ever to ship a lot of things, and a lot of bad things. Product sense and real customers are how you tell them apart.”
- Martin Knapic, Product MoatI'm really happy to host Grega this week.
We met last year at the Product Management Festival in Zurich, agreed to stay in touch, and have been exchanging PM experiences every month since. Grega is a truly dedicated PM, always hungry to stay on top of product management as a discipline. I'm impressed by how he thinks about the craft and how open he stays to learning. He recently attended a number of events in Silicon Valley and around Europe, and he couldn't hide the excitement in his voice when talking about what other companies are doing.
Grega has been helping Visionect grow since 2015. He joined as a developer, worked on Joan from its very first days as a single e-paper display next to a meeting room door, and grew with it into his role as Head of Product. Today Joan is a full workplace platform used by thousands of companies in more than 90 countries, and Grega has been part of every step of that journey.
Beyond all that, Grega has become a friend. He recently invited me to visit the Visionect offices in Ljubljana, and I was hosted at his home and got to meet his family. He's not only an extraordinary PM, he's also building a pergola entirely by himself. His DIY skills stretch effortlessly from his lovely terrace to Claude Code :)
I'm honored to have Grega on Product Moat, and I hope you enjoy reading his take on product management.
Identity & Background
- Name
- Grega Pusnik
- Current role
- Head of Product
- Current company
- Visionect (Joan)
- Location
- Ljubljana, Slovenia
- Years in product
- 7 years
Your story: how did you end up in product?
I'm an engineer. For my thesis at the Faculty of Computer and Information Science in Ljubljana I built a robot that detects and follows a person, with a Microsoft Kinect camera on an iRobot Roomba. After that I worked at the faculty for a while, then at a Slovenian startup, and in March 2015 I joined Visionect as a developer. One of my first tasks was the Google calendar integration for Joan, which back then was a small e-paper display next to a meeting room door that showed if the room was free. For the first years I was pretty much a one-man band on Joan, doing development, support and feedback gathering from the first customers. The product grew, so did the team, and I was coding less and spending more time on what Joan should be. Developer, product owner, product manager, and now Head of Product, all at the same company for more than 11 years.
Core PM Philosophy
How do you define product success?
The signal I trust most is adoption and word of mouth. I look at activation and retention first, so how many people start using a feature and how many still use it a few months later. I like to be data-driven, and these numbers should tell you if you solved a pain for users and if you need more tweaks. Revenue is important, but it comes later and you can't always directly link usage and sales. Usually it's a combination of activities and it's hard to say it was this one thing. But in the end if the product is good it will spread and grow by the word of mouth on its own. For a product to succeed, multiple teams need to work together. Product needs well written PRD and has to solve a real pain for the user, and product sense matters a lot. UX/UI needs to make it enjoyable to use, engineering needs to ship it with minimum bugs, marketing needs a compelling story, QA needs to make sure the quality is there, and customer success makes sure customers use it fully. If all this works together, the metrics will show it. Sometimes adoption is super fast after the first release, sometimes it's more of a grind figuring it out.
What's your decision-making philosophy?
I come from engineering, so I'm data driven by default. After 11 years in the same niche you do get a feeling for what will work, but I still want every hypothesis backed by data. Before we commit to something bigger, we write a Product Information Pack. It's a short template: what it is, what problem it solves, how it works, go-to-market. One of the questions is simply "What data supports this thesis?" When the data isn't complete, I try to make the decision easy to reverse. When we changed our pricing model last year, we first rolled it out in one market with the hypothesis that acquisition shouldn't drop, and only then went worldwide. Similar example when research on a new device showed we were mostly driven by FOMO, we postponed it and put the effort into an existing device instead.
What's your biggest PM lesson or mistake?
My biggest mistake was probably building product before the idea was validated. We did that more than once. An idea looked good to us, or one or two customers asked for it, so we built it and tried to sell it afterwards. Then it turned out customers didn't need it that much, and it was hard to get them to use it. But if customers need something a lot, like desk booking during COVID, and you build a great experience, adoption grows by itself. So now I want an idea tested with customers before we start building, and if we can, we try to sell it before it's built. When sales brings a feature request, I ask: would this customer sign with us if we implemented it? Is it well alligned with our north star?
How do you manage stakeholders?
Communication, communication, communication. One of the important parts of my job is making sure everybody is rowing in the same direction, towards the same north star. When people disagree, I try to bring the discussion back to data and to what customers actually told us, so it doesn't come down to whose opinion is louder.
Product & Decision-Making
How do you prioritize?
We built our own flow on top of Jira Product Discovery. All feedback goes through a single entry point, an insights mailbox. Customers, sales, customer success, insight calls, everything. Each email automatically becomes an insight ticket in Jira. The mailbox is groomed every week, and product designers and product managers take turns doing it, so everybody stays in touch with the feedback, not only the PMs. Each feature idea has its own ticket and we link all the feedback we get to it. The number of insights is one input to the impact score. The others are the expected impact on our core KPIs, like daily active users, and how much the idea would wow customers or set us apart from competitors. Effort pulls the score down. We're automating parts of this with AI, but one thing stays manual. Every product manager has to read the feedback itself, the whole insight call or the whole email, not just the summary. I believe this is important to capture the nuances of the feedback.
How do you do customer discovery and validation?
Joan itself started as an experiment. Before we built the meeting room display, we put up marketing pages and sent out emails to see if anyone cared. A lot of inquiries came in, and that's what convinced us to build it. I still like to test demand before we build anything big. The discovery round I remember best was in 2020. Nobody was in the office and our room booking device sales plummeted. We ran a series of insight calls with customers about returning to the office and working from home. The pattern was clear. We built desk booking and visitor management as software and offered it to our existing customers first. They already knew us and it cost us almost nothing to reach them. That was the start of our change from a hardware-only business to one that also sells software.
What's your favorite product you've shipped?
Joan. I've worked on it pretty much from the beginning, so it's hard to pick anything else. It started in 2015 as one e-paper display next to a meeting room door, running on battery and showing if the room is free. Then COVID emptied the offices and we had to follow how people work. Desk booking came in 2020, visitor management in 2023, digital signage in 2025. Now there's Joan AI and Joan e-paper badges for employees and events. I'm proud of where we started and where we are now. We're a small company from Ljubljana, and today Joan is used by thousands of companies in more than 90 countries.
What product inspires you?
Self-driving cars, be it Waymo in San Francisco or Tesla's Full Self-Driving. Will our children even drive on their own? This is what excites me, it's the future it will bring.
AI & the Future
How is AI changing product management?
Oh man, what isn't it changing :) It's easier than ever to ship a lot of things. And a lot of bad things as well. The core skills stay the same. Product sense, talking to your customers, figuring out what the product should do and how people will have the best experience with it. What changed is how much of the work in between I do with AI. I prepare specs with AI, but I read every line, because I'm accountable for what's in them. The roles of engineering, product and design are somewhat merging, but experts will stay. Their job moves towards writing skills others can use. Our designers write down how we design and capture it in a skill a PM can use. Engineering does the same, and with both a PM can build an 80% good enough prototype to show customers. We put it behind a feature flag and let real customers try it, without designers or engineers in the early loop. If customers confirm it, it gets merged into the core product and released to everyone. Before that, design, engineering and QA do a full review, so the final quality is spot on. We'll see how these answers age. I always tell people that the AI we use today is the worst it will ever be. From here it only gets better.
Which AI product do you use and respect?
There is a lot of them. But if I would mention one is Claude from Anthropic. As a company we're tightly integrated with it. It's connected to the tools we use every day, like Slack, Jira, Confluence and email, so it has the context it needs. I use it for first drafts of PRDs, product presentations, competitor research, and for meeting minutes that go to Confluence with the Jira tickets already created. I also built a small agent for myself. Every hour it goes through Slack, email, Jira and Confluence and prepares draft replies. Nothing is sent automatically, I review everything. All my comments, changes or fixes on its work are then used to self improve it.
What's one thing about AI everyone gets wrong?
That you spend no time on product work because AI writes all the documentation. Somebody still has to read those documents, guide the AI through your own thought process, and in the end you're the one responsible for the decisions. At a meetup in San Francisco a founder showed how an agent posted in their Slack that activation had dropped by a third. It had actually gone up. If nobody checks, a number like that goes around the company and people make decisions on it. Same goes for skills. Everyone is writing skills now, and the problem is quality and ownership. Who checks that the data in a skill is still valid? You're responsible for the skills you write and you need to keep them updated, same as code. If marketing owns the positioning research, marketing owns the skill built on it.
Beliefs & Opinions
Which author, thinker, or influence shapes you?
Lenny Rachitsky. If you haven't listened to his podcasts, you're missing out. The guests talk about what they actually do day to day.
What makes a great product?
That's really a hard one... A product you use daily and you're not able to do your work without it. It becomes the tool you go to whenever you have that specific problem. I'd add the last mile. A feature that's 99% done can feel worse than no feature, because it breaks exactly when you rely on it. If a display on a meeting room door shows the wrong status once, people stop trusting it.
Company & Product
What does your product solve?
Joan helps companies run their office: meeting rooms, desks, visitors, parking, digital signage and employee badges, in one platform. It's for workplace and facility managers and IT, and in the end for every employee who needs a room in five minutes. You've probably had these problems yourself. A room is booked in the calendar but empty, or it's free in the calendar and you walk in on a meeting. Visitors wait at reception while someone looks for their host. Hybrid work made rooms and desks harder to plan, because you never know who will be in on which day. And then there are printed employee ID cards, which are out of date as soon as someone changes team or gets a new photo. Joan connects to the calendars companies already use, like Google and Microsoft 365, and adds what calendars don't have. E-paper displays at the door that need charging only a few times a year. Booking from your phone, Teams or Slack. Check-in, so rooms nobody showed up for get released. Visitor sign-in at reception. E-paper employee badges that the office manager updates from the Joan app, so nobody has to print a new plastic card. And with Joan AI, you can also just ask an assistant to book for you.
What's the biggest misconception about it?
People think Joan is the same as the plastic Android tablets you see next to a lot of meeting room doors. Then they hold one in their hand. The case is anodized aluminium and the front is anti-reflective glass, and that makes the whole difference. The second one is battery life. People are usually surprised when they hear how long it lasts. E-paper uses power only when the image changes, so Joan 6 Pro lasts up to six months on one charge and Joan 13 up to a year. And because it runs on battery, there are no cables to the door, so no electrician and no drilling through walls.
Why would you recommend it to a peer?
If you run an office or support one from IT, take a look. You get one platform for rooms, desks, visitors and signage instead of separate tools, and devices that don't need any cables. And if you're a PM, Joan is an interesting case study. Hardware and SaaS in one product means two very different release cycles that have to fit together. Happy to talk about it, just reach out on LinkedIn.
More from the conversation
What would you tell a PM just starting out?
Do support for a while, even if it's not your job. Answering tickets about code I wrote myself taught me very fast what customers actually care about. You need to deeply care about the product, care about your customers. This is what it will build your product sense. Go to meetups, talk with your peers, discuss about the problems you have. You will be able to progress much faster. And learn to build something yourself. With the AI tools today it's not that hard anymore, and customers give you much better answers when they can click through something than when you describe it to them.
Enjoyed this conversation?
A new interview with a product person lands every week. Get it in your inbox.