Interactions
with AI.

A Teams attendance report. A Word document. Copy a name, paste a name, go back for the next one. Repeat after the next meeting. You recognised how uncomfortable that gets almost immediately.

Karin, one of the executive assistants in the Volvo Group Next Step programme, tried something else. She gave Copilot the whole report and asked for a list containing just the names. One interaction replaced a sequence of repetitive steps.

Then she shared it. A colleague wanted an Outlook-ready list of email addresses. They tried that too. The instructions became reusable, the approach spread, and the work moved towards an agent handling the recurring task.

The important change was in who could act. Someone who had spent years following other people's digital instructions could now change the instructions herself. And help her colleagues do the same.

I brought up my runner-up experience in the Swedish prompting championships, then explained why I now spend so much of my time talking with AI. A conversation gives us room to discover the question, supply missing context and change direction.

Careful instructions still earn their place. Karin's example contains both modes: explore together until you understand the operation, then capture the instructions so it can be repeated. The useful skill is knowing when to specify and when to explore.

The planned live voice exchange with Petty ran into an audio problem. That demonstration did not work in the room. The invitation still stands: try explaining a real problem out loud, or use the same conversational approach in text. You do not need a stage setup to begin.

The Kasparov story moves us through fear, relief and responsibility. A machine can outperform extraordinary human skill. Combining human and machine capabilities creates another possibility. But merely putting the two beside each other does not make a good team.

My three-law teaching story is an interpretation of Kasparov's work. Its practical landing is to choose an appropriate tool, learn how to interact with it and build a process that holds up in use. As I also said on stage, adding a human does not automatically improve a machine's decision.

Kasparov's published account of freestyle chess makes the process especially clear: amateurs using a better way of working with their computers overcame stronger players and machines. That is a story about collaboration, not a universal performance guarantee. Read his account.

My AI team had given me a poor morning preparation. I got annoyed. It pushed back: it had been asking me how to handle the gaps, and I had not answered.

The label got my attention. What mattered was the conversation that followed. What did the evidence say? Which failures belonged to me, which to the system, and what would actually change how we worked?

During my years in the army, structured feedback was part of the work: upwards, downwards and sideways. I have missed that. In this exchange, I recognised something worth developing again.

The next part matters just as much: keep talking about how to improve. A harsh verdict is not automatically useful. Ask for the observation behind it, examine the interpretation and agree a change you can test.

I also demonstrated asking for feedback from selected computer activity and accumulated context. The response suggested that I delegate completion more clearly and spend less time coordinating the work myself. That was an interpretation to discuss, not an objective personality assessment.

This is what I mean here by mutual growth: I can change my habits, and the shared instructions, context and working process can improve. It does not require the AI to experience the relationship as I do.

Karin's story includes time saved. It also includes agency, learning, confidence and shared capability. The keynote put particular emphasis on access and inclusion: who gets to write, analyse, build and contribute when a previous barrier becomes easier to cross?

These six value vectors help widen the conversation. Choose the ones that matter to your work and look for evidence that something improved.

An ongoing working relationship needs context. What matters to you? What does good work look like? How do you prefer to be challenged? What may your collaborator decide without asking?

My public profile is a deliberate account of who I am and how I work. It includes life, values, learning and working preferences. It makes no claim to be an independent, objective portrait. You and your AI can read it, ask better questions and form your own judgement.

In the Lagunen demonstration, I compared a reply to a cold enquiry with one that could also draw on my public context. The point was to show how an interaction could become more relevant before we had met. It was an illustrative demo, not a real booking or a measured business result.

PPCP, the Public Personal Context Protocol, is the open proposal behind this way of sharing personal context. Private working context and public disclosure are separate choices. Start with what you actually want another person to know; publishing a profile is optional.