For most of my early years at Hellon a project lived on a wall. We printed the interview quotes, cut them into strips and moved them around until a pattern held. The kick-off notes were in one person's notebook and the client's old strategy deck was on someone's laptop. Decisions were made standing up, with a marker in hand, and the wall remembered them for us.
I liked it, and I still think it produces some of the best thinking. It also meant the project was held by the three or four people who had been in the room. When one of them moved to another project, the rest of us asked around until we found what we were looking for, and most of the time we found it.
That way of working suited the amount of material a project used to carry. A client engagement today brings in a broader context; more interviews, more documents and more history than a wall can hold or a team can remember, and the thinking has to happen somewhere that can hold all of it.
Then AI arrived, for all of us at once. Within a year everyone at Hellon had an assistant open in a browser tab and used it on their own, with whatever they had pasted in that morning. I did the same. Research summaries got faster and first drafts came together in an afternoon.
The frustrating part was what happened between people. I would work through a question with the AI, bring the result to the team, and find that a colleague had done the same thing from a different set of material and arrived somewhere else. We spent the meeting reconciling two versions of the same problem. Each assistant knew only what its owner had told it, so the gaps between us were wider than they had been in the sticky-note days, when at least we had all been looking at the same wall.
This is how a project of mine runs now, inside what we call AINO OS.
It begins with the proposal we have agreed with the client. I open a project for it in Claude and ask it to find everything relevant that the proposal leaves out: earlier work with the same client, the conversations we had while preparing the offer and my own notes from them. Claude reads across our meeting transcriptions, our internal chat and our file database and compiles a context document, which I load into the project. This is the knowledge that used to sit in people's heads and in scattered threads, and it is often what decides whether a project goes well.
The context also carries our Quality Framework, so the project is guided by it from the first day rather than checked against it at the end. Claude keeps the framework in view in two ways. It reminds us of the three levels as we go, the foundation we must deliver, the value we can add and the extra that makes the work memorable, and it makes those levels specific to this client: what the foundation deliverables are here, what needs tailoring to this organisation, where we could plan for something they have not asked for. Claude widens what is on the table and the team decides what reaches the client.
As the weeks pass, Claude's picture of the project keeps up with it. Each week it gathers what happened across our discussions, notes and tools, what we delivered and what we planned, and gives us an outlook for the week ahead. It also sees my calendar, so I arrive at meetings prepared. Workshop preparation is the ordinary example. Claude lays out the goals and a first structure before we meet, the team and the client reshape it, and from there it builds the agenda, the exercises and the boards as a ready draft for my review.
Most of the day-to-day weight of running a project has moved off my desk. What is left is the strategic, creative and analytical work I need to do with my team, and there is more time for it than before.
A recent project showed me what this adds that the old way could not. The client had come to us with an initiative that kept stalling, and the brief was to work out why and get it moving. Our team was puzzled, why do we hit a wall all the time? A few years earlier Hellon had worked with the same organisation on a different question, with a team that has since moved on. Nobody on the current project had been part of it.
While compiling the context in the first week, Claude surfaced a decision from that earlier engagement. Read against the new brief, it explained a good part of the stalling. The organisation was trying to move in a direction it had, in effect, ruled out some years before, and the people now in charge did not know that.
We would have found this only if someone from the old team had happened to remember and happened to be nearby. Instead we had it sent to us by Claude doing the digging. The team spent a long afternoon working out what it meant for this client and how to raise it with them, and that afternoon changed the shape of the whole engagement.
Claude found the note, but the team decided what it meant and what to do about it. Now this is happening all the time, across the organization.
It would be easy to read all of this as handing the work to AI, but that is the opposite of what happens. Because the gathering, the first drafts and the weekly bookkeeping are off my desk, the team spends its time where human collaboration pays: in the room with the client, at the wall, making sense of things together. We still print things out. We still argue about what a pattern means. But we do it with more material in front of us and more hours to spend on the part that matters. This is why I have started calling AINO OS a collaboration machine.
The same system is also how we keep the work safe. The Quality Framework names what must be right before anything leaves us: facts and sources that are traceable, the brief held in scope, the client's context reflected, and a named person who has signed the work off. Our guardrails set which tools may touch which material, what we check with the client, and what stays outside AI altogether.
None of this arrived from nowhere. In 2017 our data scientists added machine learning to our segmentation work, which for the first time let us combine what we had heard from customers with what the numbers said, in the same analysis. That was one piece, put in place years before large language models were published. AINO OS is the same instinct with far more reach: give the people in the room more to think with, and leave the thinking to them.
The film at the top of this post ends with the words human to AI to human. That is the order I would defend. The AI sits in the middle, and the work still begins and ends with people in a room, which is how business has always worked at Hellon.
Authors:
Juha Kronqvist is a strategic design consultant and a pioneer in the Scandinavian market, with a proven track record of delivering innovative service concepts through deep understanding of markets, audiences, and their needs. He possesses industry-leading expertise in applying cutting-edge methodologies to achieve groundbreaking results.