Separating Noise from Reality at Constellation Software: A Portfolio Manager’s Insights on Customer Focus and AI
In this conversation, Kanchan Java, President of the Pegasus Portfolio at Perseus Group, shares with Pavel Prokofiev and the Podcast Community of RollUpEurope on how Constellation operators approach integration, growth, and innovation in a highly decentralised environment while drawing on her experience leading Charter Software through a period of significant transformation.
You can listen to the conversation here:
1. Why onboarding new acquisitions at Constellation isn’t as cookie-cutter as you might think
Pavel: When you acquire a new business, what does onboarding look like? Do you get a playbook, or is it more a case of good luck figuring it out yourself?
Kanchan: I wish there was a document that everyone worked from in the same way. However, since Constellation is very decentralised, we have collections of playbooks in different places. Playbook number one is the integration plan, which we work on before we even close the acquisition.
Pavel: When you buy a new business, how much pressure is there to hit financial targets in year one versus focusing on fixing the foundation first?
Kanchan: We evaluate a lot of our businesses on IRR, and anyone familiar with IRR knows that the timing of cash flows really matters.
In mature vertical market software businesses, you can continue to run profitably while reinvesting into the product, into customer satisfaction, into future growth.
Pavel: What happens when things don’t go according to plan? Say you’ve executed a price increase or an operational improvement, and it’s not landing the way you expected.
Kanchan: You go back to the investment case and re-evaluate. With Charter, we had competing products in our portfolio and weren’t sure how to support them all long-term.
Our concerns proved to be unfounded since. Charter catered to smaller agricultural dealerships and outdoor power equipment dealers, while another brand in our portfolio served large agricultural dealerships. There was room for both.
Once we confirmed that Charter had a strong niche and that the customers for one product weren’t interested in our other products, we pivoted. We decided to grow the business, reinvest seriously into the product, and that’s when I got involved.
2. Post acquisition: a road trip to see 50 customers
Pavel: In your first year, you visited 50 customers. Walk me through what those visits looked like, and what you were watching for.
Kanchan: We planned it as a series of road trips: we’d fly into one city and out of another, fitting in as many customers as possible along the way.
The first thing we did was map all our customers geographically – you’d be amazed how often people focus only on large accounts and miss a smaller customer 90 minutes down the road.
We would spend a couple of hours at each customer’s site. The goal we shared in advance was simple: we want to understand where your business is headed. What are your pain points? What are your opportunities? Understanding what customers are trying to achieve means you can align your roadmap with them rather than just collecting a wish list.
We’d talk through our product roadmap. If you don’t proactively tell customers what you’re building, they assume the list of everything they want is still open. You consolidate 10 requests from Customer A and 10 from Customer B into something coherent, rather than drowning in competing priorities.
We brought trainers on some visits. Customers got some free training, and we got to watch real users interact with the software. Hearing directly from the people using the product every day is incredibly valuable.
Sometimes, I took developers along. That was eye-opening. They’d be sitting there thinking about complex, technically interesting problems to solve – and a customer would say, ‘When I get to this field, the cursor ends up over there, and every single day I have to move it before I start typing. Can you just make it start in the right place?’. This is the stuff you cannot get from a phone call or a feature request form.
We’d look around. Is there a whiteboard on the wall? What are you capturing there that our software doesn’t handle? Are you jumping from our tool into another tool? Can we integrate better or bring that functionality in-house? It’s the kind of insight you only get by being in the room.
3. Why you shouldn’t have R&D running Product
Pavel: After the customer visits, what were the first hires or structural changes you made?
Kanchan: Something that was stressed to me very early by experienced GMs within Constellation: separate product management from R&D.
In a lot of small businesses we acquire, R&D decides what goes into the software. It’s the developers, or in many cases the founder themselves, who talks to customers and then comes back and says, ‘We’re building this and this.’ If the founder retires after the sale, you have a complete void. If the R&D person is filling that role, they’re often building things customers don’t end up using, because developers don’t have much face time with customers.
One of the first meaningful changes we made at Charter was hiring a product manager. She came from outside the industry. Our customer surveys showed the software wasn’t easy to use. It struck me that the existing team couldn’t see it. They thought it was a training issue. This happens a lot with products that have been built over 10, 15 years by responding to customer requests. You end up with enormous functionality, but no one’s been thinking about UI/UX or whether it’s intuitive.
The product manager brought a fresh perspective. She’d never seen the software before. She was a problem solver who liked working with rural customers – exactly our customer base. We gave her a clear brief: here’s the problem, make it easier to use.
While watching the trainers, she kept hearing things like: ‘I know it says this on screen, but what it actually means is this.’ So, she asked: why not just rename the fields so they say what people expect them to mean? Internally, there was pushback – developers thought renaming fields was a waste of their time when they had important work to do. I said there is nothing more important than being intuitive to the customer.
4. What does firmwide AI adoption look like in a highly decentralised organisation like Constellation
Pavel: After the successful integration of Charter you were promoted to a portfolio President. Constellation is actively embedding AI into products – computer vision, production issue flagging, plain-language data queries. Can you pick one of those initiatives and walk me through the mechanics and the customer ROI?
Kanchan: Let’s take computer vision. This came out of a partnership: we met a company with a relevant product, we had a customer interested in it, so we said – let’s work together. That is how I think about AI strategy: figure out what you can build, what you want to partner on, and what you want to buy. With the pace of technology evolving, trying to build everything yourself is a mistake.
We don’t want to use AI just to say we use AI. The question is: what problem can we now solve that we couldn’t solve before?
For example, we’re upgrading our Manufacturing Execution System (MES) to allow customers to use existing cameras in their facilities to detect issues on the manufacturing floor. Something as straightforward as an employee not wearing personal protective equipment, or a fault on a paper roll. We have many customers in the pulp and paper industry. The AI flags these events to a human operator in real time.
Pavel: Is there a Constellation-wide AI strategy, or is each general manager making their own bets?
Kanchan: What I can tell you is that our senior leadership is actively working to embed AI into products and on how to improve operational efficiency with AI. A lot of them have development backgrounds, and they’ve validated firsthand that these tools do make it easier to develop.
All of the businesses I oversee came into 2026 strategy planning with AI as a top priority. At the individual business level, people are working independently – experimenting, tinkering. One group started using GitHub Copilot, found they preferred Cursor, switched, and then put together a presentation explaining why. Several other companies immediately tried Cursor too.
The technology is moving so fast that I see tremendous advantage in people exploring independently and then sharing what works. We don’t mandate a single tool – we share best practices and let people make informed choices.
Pavel: Have you already seen measurable improvement in developer productivity from AI tools?
Kanchan: Yes we have. We have run controlled experiments – estimate how long this task would take with your old tools, now do it with the new ones – and consistently seen significant productivity improvements in those tests. The investment goes back into the product.
5. What does “decentralisation” actually mean to Kanchan? And what are her priorities?
Pavel: You mentioned Constellation’s decentralised model a few times. What does that mean to you personally?
Kanchan: When I joined Constellation, I had a background in M&A, and in every previous deal I’d worked on, the goal was always to blend the acquired business with something we already owned. Synergies was the word – one management team, one process, run both businesses together more efficiently.
When I joined Constellation, I was genuinely surprised that we don’t do that. When we buy two businesses that do similar things, we look at them and say: there’s room for both in this market. And we keep them separate.
You give up something in cost synergies, maybe. What you get in return is closeness to the customer and closeness to the team. That closeness makes you agile and able to do genuinely great things. Bureaucracy, in my view, is the enemy of innovation and creativity. What we’re betting on with our model is that the people closest to customers make the best decisions – for their customers, and for growth.
What we add is information. Running a small independent software business can be lonely. You sit there wondering: am I spending the right amount on R&D? Should I be investing more in sales and marketing? How do I move the business forward? There’s no one to ask. Maybe your spouse. Maybe a cousin. Within Constellation, you can talk to people in other businesses, tap into portfolio-wide knowledge, and get connected with someone who has faced your exact challenge. Someone moved to the cloud last year? Let me introduce you. Conversations like these create value.
A lot of our general managers are inherently competitive and inherently curious. They love seeing what others are doing, learning from it, and incrementally improving their businesses every year – on their own terms, without being told exactly how.
Pavel: Looking ahead – what word would you like to get out to our readers? What are you looking for?
Kanchan: Our model is simple: buy great businesses, empower teams, share knowledge and playbooks, and let people run. The best thing we can do as we continue to acquire is hire future leaders and give them real opportunities to grow into roles.
Back to Insights