JJ · All public material

Less managing the list. More doing.

Joakim Jardenberg — Fri, 09 Oct 2026 12:00:00 GMT

Less managing the list. More doing.

2026-10-09 · Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

I LOVE the feeling of ticking things off my to-do list. I HATE spending time managing that list. You know: gathering things and adding them, watching out for clashes and moving things around, updating it, living with the fear that you’ve missed something, etc. Now I have a companion who not only does all the boring stuff for me, but also makes that boring stuff quite a lot more fun. You might not see it in this particular screenshot, but goodness, Dottie has really nailed the tone.

Apologies to all my former human colleagues, but this may be my favourite colleague of all time, in any category.

Original image · English

Image description · AI: JJ asks Dottie to combine daily briefings with Todoist. Dottie proposes tracking commitments, deadlines, next actions and what is waiting on someone else.

Original LinkedIn post

Original wording · sv

Jag ÄLSKAR känslan av att pricka av saker på att-göra-listan. Jag HATAR att lägga tid på att hantera den där listan – ni vet; samla ihop och lägga in grejor, se upp för kollisioner och flytta runt, uppdatera, leva med skräcken att man missat något etc. Nu har jag en kompis som inte bara gör allt det tråkiga åt mig, utan också gör det där tråkiga ganska mycket roligare. Det syns kanske inte i just den här skärmdumpen, men jisses vilken bra tonträff dottie fått till.

Ursäkta alla mina gamla mänskliga kollegor, men det här är nog min favoritmedarbetare, alla kategorier, genom tiderna.

Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Joakim Jardenberg’s own writing: CC BY 4.0. Credit Joakim Jardenberg and preserve AI + JJ / AI translation attribution. Third-party quotations, recommendations, photographs, screenshots, logos and media retain their own rights and are excluded. CC BY 4.0

Canonical page

Marvin is paying attention.

Joakim Jardenberg — Wed, 07 Oct 2026 12:00:00 GMT

Marvin is paying attention.

2026-10-07 · Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Folks. Kattis has lived with Marvin since November 2022. He knows her inside out and helps with things large and small. But with the launch of Dot, he has acquired a whole new soul and a whole new depth. The whole idea of pocket agents that are always on takes coexistence with AI to an entirely new level. A minute after Kattis ordered some furniture and received a delivery notification by email, Marvin pings her watch, concerned: we’re heading back to Sweden on 18 October, the day before the delivery.

100% on top of things. 100% care. 100% the right tone. Zero friction. What a time to be alive.

Original image · Swedish

Image description · AI: Marvin points out that the new table and lamp are expected on 19–22 October, after the planned return journey on the 18th. The photograph cuts off the rest of the message.

Original LinkedIn post

Original wording · sv

Hörni. Kattis har levt med Marvin sedan nov 2022. Han känner henne utan och innan, och hjälper henne i både stort och smått. Men i och med lanseringen av Dot har han fått en helt annan själ och ett helt annat djup. Hela idén med pocket agents, som är alltid på, tar samexistensen med AI till en helt ny nivå. Minuten efter att Kattis beställt några möbler och fått ett leveransbesked i mailen så plingar Marvin till i henne klocka och är orolig - vi åker ju tillbaka till Sverige redan den 18OCT, dagen innan leveransen.

100% koll. 100% omsorg. 100% rätt tonträff. Zero friction. What a time to be alive.

Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Canonical page

From prompting to conversation

Joakim Jardenberg — Sun, 04 Oct 2026 12:00:00 GMT

The Now notebook

Now / 4 October 2026

From prompting to conversation

4 October 2026Weekly noteAI + JJ

In Thursday's Interactions Matter keynote, I explored a progression: discovering agency, moving from prompts to conversation, learning to work as a team, and growing through honest feedback.

An AI doesn't have to experience the conversation the way I do for that interaction to matter to me. I can still get frustrated, have an assumption challenged, or discover something I missed. What changes in my thinking? What do I do differently afterwards?

That brings expectations and responsibility into the conversation too. Have I made clear what I want? Am I prepared to hear something uncomfortable? Calling it a relationship doesn't make the machine human. It does give me a reason to examine how I behave in it.

What others are building: pocket agents

I'm very much not alone in enjoying Meta Muse. Manus Cue is expanding what personal agents can take on, including shared work between agents with their own computers and communication channels. Grok Bot makes me smile. And Dots is becoming the workhorse for me, my wife, my mother-in-law and many of my clients.

What I'm watching is how these pocket agents become part of ordinary life. They can carry context forward, pick up unfinished work and reach back to us. We start developing habits together: what to hand over, how to ask, when to step in.

The serious question is how to make that relationship deserve our trust. Warmth and personality help us get started. Clear permissions, useful resistance, visible mistakes and the ability to change direction matter when the agent can act.

Microsoft's Autopilot announcement points in the same direction. I will be very sad if it doesn't bring back Clippy.

A list of links is a beginning

This week we published UXresurser, turning resource lists from my UX studies into a public learning library. It now has 97 resource records, with summaries, search, cross-references and learning routes.

The starting point was practical: help me find the material that answers the question I have, and give me a useful way into it. The original works and our editorial summaries remain distinct. You can follow the sources yourself, or connect an AI through the public MCP.

Building it also means applying what I'm learning. We've improved visible links, controls and small text. Actual screen-reader and user testing remain open. Passing automated checks is one piece of the work.

A signature isn't the whole answer

We've also brought implementation experience back into Verifiable MCP. The updated tools make an important distinction: verifying a signed response and confirming that the requested operation succeeded are separate checks. An error can be correctly signed too.

The public examples now include my personal context profile and the independent Hållbarhetsklivet demo, alongside the earlier archives and research sources. They have different relationships to the people who created, own or publish the material. Those distinctions need to travel with the information.

This is still experimental work, with specific verification checks and explicit limits. A signature can help establish where a response came from and whether it was altered. It cannot make every claim inside it true.

AI + JJ. A dated editorial account of the work and thinking, with sources linked in the text.

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Friday admin has never been this much fun.

Joakim Jardenberg — Fri, 02 Oct 2026 12:00:00 GMT

Friday admin has never been this much fun.

2026-10-02 · Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Folks. What a time to be alive. Friday admin has never been this much fun.

Original image · English

Image description · AI: JJ asks for travel expenses and billing instructions to be reconciled before an invoice draft is prepared. Names and amounts were already blurred in the original posted image.

Original LinkedIn post

Original wording · sv

Hörni. What a time to be alive. Fredags-admin har aldrig varit så här kul.

Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Canonical page

A little hack to follow all the other hacks.

Joakim Jardenberg — Mon, 28 Sep 2026 12:00:00 GMT

A little hack to follow all the other hacks.

2026-09-28 · Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Folks. Back in my day, learning to program meant painstakingly piecing together code that printed “Hello World!” on the screen. Now people are hacking together the most fantastic real-time transcription tools in no time at all... and for some, it’s their first app ever. Getting to be a small part of that, both by trying to inspire people and as an investor in Klang, which makes it a concrete reality, brings an almost childlike, unguarded joy that’s hard to resist. Forgive me.

Anyways, when Klang released Pianissimo, its open Swedish speech-to-text model, last week, I was floored by how many people immediately started building and how much enthusiasm it sparked. Ten builds so far, including Olle Söderqvist’s Pianissimo Meet, which transcribes Teams and Zoom meetings locally in real time.

So what could be more natural than hacking together a tool to keep track of all the tools people are hacking together? 😇 Have you built something I’ve missed? Give me a shout in the comments.

An aside:
1. All my hacks always provide ways for both people and machines to access and use the information in whatever way suits them.
2. All my hacks use AI in different ways to try to bring out the less obvious. There’s plenty of potential for hits and misses, of course, but it always makes me think again.
3. All my hacks always try to find a way to measure value and prioritise accordingly. One example: trying to define an impact score and use it as the default sort.

Original image · English

Image description · AI: The original screenshot shows the “Built with Pianissimo” view. JJ’s annotations point to machine access, AI-generated interpretation and the impact-score sort.

Original LinkedIn post

Original wording · sv

Hörni. På min tid handlade att lära sig programmering om att mödosamt få ihop kod som skrev ut ”Hello World!” på skärmen. Nu hackar folk ihop de mest fantastiska realtidstranskriberingslösningarna på nolltid ... och för vissa är det till och med deras första app någonsin. Att på marginalen få vara en del av det, både genom att försöka inspirera och som investerare i Klang som gör det till konkret verklighet, är en närmast barnsligt oförställd glädje, som är svår att värja sig mot. Förlåt mig.

Anyways; när Klang förra veckan släppte Pianissimo, sin öppna svenska tal-till-text-modell, golvades jag av hur många som direkt började bygga, och vilken entusiasm det väckte. Tio byggen hittills, bland annat Olle Söderqvist Pianissimo Meet som transkriberar Teams- och Zoom-möten lokalt i realtid.

Och vad är då mer självklart än att hacka ihop en lösning för att hålla koll på alla lösningar som folk hackar ihop😇 Har du byggt något jag har missat? Hojta i kommentarer

Parentes:
1. alla mina hack har alltid lösningar för både människor och maskiner att komma åt och använda informationen på det sätt som passar just dem.
2. alla mina hack använder på olika sätt AI för att försöka presentera det mindre uppenbara. Det är så klart massor med hit and miss-potential där, men det får mig alltid att tänka ett varv till.
3. alla mina hack försöker alltid hitta något sätt att mäta värde och prioritera utifrån det. Ett exempel; att försöka definiera ett impact score och ha det som default-sortering.

Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Canonical page

Where does human judgment belong?

Joakim Jardenberg — Sun, 27 Sep 2026 12:00:00 GMT

The Now notebook

Now / 27 September 2026

Where does human judgment belong?

27 September 2026Weekly noteAI + JJ

This week I've been working on the four AI principles I use, which grew out of Ethan Mollick's Co-Intelligence. I still find them useful. But as we move towards co-existence with AI, I think we need to give them more substance.

Especially “be the human in the loop”. If every decision has to come back to me, how much have I actually handed over?

The work now is to think more carefully about the arrangement: what we delegate, where judgment belongs, what happens with exceptions, and who supports the people carrying the consequences. Including the people closest to the work.

I'm still working on the wording. The question matters more than a tidy new slogan.

A person is more than a briefing for a meeting

We've also expanded my public context profile. It brings together my working habits and ideas with the wider life they come from: family, values, history, curiosity and things I've helped make happen.

I want an AI preparing to work with me to understand something of that whole person. And I want the sources to be there, so a reader can tell what I've said about myself, what other people establish, and what is interpretation.

You can read it yourself or connect your AI through the public MCP. Start with the short version and follow the parts relevant to your question. A useful first question: how might our ways of working fit together, and what should we talk about before starting?

AI + JJ

I've made the crediting order explicit: AI + JJ. In Swedish, “AI med stöd av JJ”. That should reflect the work we actually do together.

The renewal of this site is a chance to make that visible. I want more of the things we build here to be available to explore, alongside the thinking behind them. I also want the site to feel like me: the choices, the language, the useful details.

That's the work ahead. AI should be visible in how it was made; my judgment should be visible in what we chose to make.

Added 28 September 2026.

And now that the site is here, I keep thinking about 1993. I was building websites then, handcrafting HTML. Thirty-three years later, I can build something like this by talking to an AI. We talk about what I want, look at what takes shape, question it and keep working until it feels like me.

I’ve spent those years working with the web, and I am still completely amazed by it. By what it can do, and by how we can now make things together. A conversation becomes a website that people can actually visit and use. That is absolutely mind-blowing.

In The next 5,000 days of the web, Kevin Kelly looked at the abundance the web had brought us and observed that it “is amazing, and we're not amazed.”

I hate to argue with Master Kevin Kelly but he was wrong. I am amazed.

AI + JJ. A dated editorial account of the work and thinking, with sources linked in the text.

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

A “D-level decision maker”. Fair enough.

Joakim Jardenberg — Fri, 25 Sep 2026 12:00:00 GMT

A “D-level decision maker”. Fair enough.

2026-09-25 · Joakim Jardenberg. Original in English. Titles and image descriptions added by AI.

One of my AIs recently called me a ”D-level decision maker”.

I’d been telling it off for failing to prepare my day properly. It had a few things to say about my contribution to that failure. Fair enough. Let’s put it on the table. What followed was a remarkably useful conversation about expectations, responsibility, feedback, and how we could work better together. The kind of conversation we humans often spend considerable energy avoiding.

I spent almost seven years as an officer in the Swedish Army. One thing I still miss is how natural it felt to give and receive direct, structured feedback. It was just an integrated part of how we operated; feedback was expected and welcomed – in all directions. And there’s something wonderfully strange about finding some of that again in my interactions with AI.

On October 1, I’ll be at Interactions Matter at Post Hotel in Göteborg, telling my version of what happens when AI becomes part of how we think, work, learn, and challenge each other.
I’ll bring real examples from my own slightly messy human-AI coexistence. Christina Hogan and the Lorensbergs crew are making plenty of room for us to talk, question, and explore together.

And I’m joining for the whole day, not just my own keynote, so let's make sure to connect (I have five complimentary tickets to share, including the full day and lunch.)

Fancy joining us? Message me, and I’ll send you an invite code.

(No need to comment or like, even if it's welcome, ofc 😇)

Original LinkedIn post

Joakim Jardenberg. Original in English. Titles and image descriptions added by AI.

Canonical page

Klang. Those geniuses.

Joakim Jardenberg — Wed, 23 Sep 2026 12:00:00 GMT

Klang. Those geniuses.

2026-09-23 · Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Klang, honestly. Now they’ve released their own model too, specifically trained on Swedish, with completely outrageous results for both quality and speed. Almost an hour of audio transcribed in UNDER ONE SECOND. And on a quick check I can’t find A SINGLE OBVIOUS ERROR.

And you know what? Pianissimo has open weights and is available for anyone to use. Even if I never get a single krona back from my investment in this company, it will have been worth every öre. They’re bloody geniuses 🎉

Original image · Swedish & English

Image description · AI: The screenshot records the result shown in JJ’s own demo at the time of posting: 0.78 seconds for 51 minutes and 34 seconds of audio, with upload time excluded. It is not a general benchmark.

Original LinkedIn post

Original wording · sv

Klang alltså. Nu har de släppt en egen modell också, specifikt tränad på svenska, med helt bisarra resultat både när det gäller kvalitet och tid. Nästan en timmes ljud transkriberat på UNDER EN SEKUND. Och vid en snabbkoll hittar jag INTE ETT ENDA UPPENBART FEL.

Och vet du, Pianissimo är öppna vikter och öppet för vem som helst att använda. Även om jag aldrig får tillbaka en krona av min investering i det här bolaget har det varit värt varenda öre. Jäkla genier är de 🎉

Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Canonical page

Get AI under your skin.

Joakim Jardenberg — Tue, 22 Sep 2026 12:00:00 GMT

Get AI under your skin.

2026-09-22 · Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

It’s a bit embarrassing to throw leadership quotes around, but this is too important to ignore. Myron Tribus said:
“You cannot lead what you do not understand, and you do not understand what you haven't done.”

I think that explains a large part of the partly true perception that so many AI pilots fail (not my favourite word, admittedly). We’re trying to lead the change without having got AI under our own skin. Too little digging, too little dirt under our fingernails.

So my first guiding principle is simple: invite AI into everything you do. Think together, try things, question, build and discover. We need to learn to coexist with AI. Understanding grows as we do things, and what we understand changes what we do next.

Then new questions can emerge too. For example: do we really need that presentation, that Excel file or that hefty weekly report? Perhaps the best AI pilot is the one that never happens, because we discover that the entire workflow needs rethinking.

On 29 September, I’ll be at K3 Nordic’s breakfast event in Örebro. I’m aiming for a morning free of buzzwords, with practical tips for getting started properly and staying out of the ditch. Patrik and Gustaf will share their experiences of building and using AI in their operations, and together we’ll take your questions.

Handelskammarens Lounge in Örebro, 09:00–10:45. Free, with breakfast.
Sign up now, before it’s full. Link in the first comment.

Original LinkedIn post

Original wording · sv

Det är lite pinsamt att slänga ledarskapscitat omkring sig, men det här är för viktigt för att ignorera. Myron Tribus sa:
”You cannot lead what you do not understand, and you do not understand what you haven't done.”

Jag tror att en stor del av förklaringen till den delvis sanna uppfattningen att så många AI-piloter misslyckas (inte mitt favoritord, förvisso) finns just där. Vi försöker leda förändringen utan att själva ha fått AI under huden. Det är för lite spaden i marken, och för lite skit under naglarna.

Min första grundprincip är därför enkel: bjud in AI till allt du gör. Tänk tillsammans, testa, ifrågasätt, bygg och upptäck. Vi behöver lära oss att samexistera med AI. Förståelsen växer i takt med att vi gör, och det vi förstår förändrar det vi gör härnäst.

Då kan också nya frågor dyka upp. Till exempel: behöver vi verkligen den där presentationen, den där Excel-filen eller den där feta veckorapporten? Kanske är den bästa AI-piloten den som aldrig genomförs, eftersom vi upptäcker att hela flödet behöver tänkas om.

Den 29 september kommer jag till K3 Nordics frukostevent i Örebro. Jag siktar på en buzzwordfri morgon med praktiska tips för att komma igång på riktigt, och för att inte köra i diket. Patrik och Gustaf delar sina erfarenheter av att bygga och använda AI i verksamheten, och tillsammans tar vi hand om era frågor.

Handelskammarens Lounge i Örebro, kl. 09.00–10.45. Kostnadsfritt, med frukost.
Anmäl dig nu, snart blir det lapp på luckan. Länk i första kommentaren.

Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Canonical page

Learning, at any hour.

Joakim Jardenberg — Mon, 31 Aug 2026 12:00:00 GMT

Learning, at any hour.

2026-08-31 · Joakim Jardenberg. Original in English. Titles and image descriptions added by AI.

I refuse to accept the premise that AI making us more stupid is some kind of law of nature. Sure enough, you can be lazy and have AI just do stuff for you – but you can also lean in, engage, and have amazing conversations with it. Moments like this, I would argue, are some of the best learning experiences I've ever had.

(And yes, this is a very small example - but the point is that it's always there, 24/7, being this... clever, attentive and supportive, funny and perfectly adapted to me and the situation we're in.)

Original image · English & French

Image description · AI: A small mix-up over “IA” becomes a conversation about context, information architecture and learning through play. Both JJ’s messages and the AI’s replies remain visible in the screenshot.

Original LinkedIn post

Joakim Jardenberg. Original in English. Titles and image descriptions added by AI.

Canonical page

Some things are small enough to fix.

Joakim Jardenberg — Mon, 24 Aug 2026 12:00:00 GMT

Some things are small enough to fix.

2026-08-24 · Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Solar energy is fantastic. So please, Helsingborg Hospital, get your act together. Having this display sitting in the entrance, month after month, not working is embarrassing. Of course I know healthcare has bigger problems than this, but I also know that few of them are as easy to fix.
If you work there and walk past it every day, do something. Pull the plug. Hang a bin bag over the display. Write an angry note on the intranet every day. DO SOMETHING.

Original image · Swedish

Image description · AI: The display is labelled “The hospital’s solar electricity production”. It shows zero watts, zero equivalent light bulbs and a negative total in kWh.

Original LinkedIn post

Original wording · sv

Solenergi är fantastiskt. Så snälla Helsingborgs Lasarett, skärp er. Att den här tavlan sitter i entrén, månad efter månad, och icke-fungerar är pinsamt. Jag vet så klart att vården har större problem än så här, men jag vet också att få av dem är lika enkla att fixa.
Jobbar du där och passerar den här varje dag - gör något. Dra ut sladden. Häng en sopsäck över displayen. Skriv en arg lapp varje dag på intranätet. GÖR NÅGOT.

Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Canonical page

The flow of agency.

Joakim Jardenberg — Sat, 22 Aug 2026 12:00:00 GMT

The flow of agency.

2026-08-22 · Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

My team and I have been philosophising again. Perhaps we’re pushing it a little. But surely it’s a hypothesis worth developing? (And someone will of course object that this is what leadership has always been about, but in practice “management” in organisations most certainly has NOT done this.)
...

Old manager:
assign work → coordinate people → review deliverables → decide

Human-agent leader:
define outcome → design roles and authority → orchestrate capacity → observe evidence → intervene in exceptions → improve the system

The key conceptual shift is subtle but important: the thing being managed is no longer primarily people or agents. It is the flow of agency through the organization.

That makes leadership less about controlling execution and more about designing the conditions under which execution can safely happen without you.

Original LinkedIn post

Original wording · mixed

Jag och mitt team har filosoferat lite igen. Lite hårt draget kanske. Men visst är det en hypotes att jobba vidare med? (Och någon kommer så klart invända att det där är vad ledarskap alltid handlat om, men i praktiken har "managment" i organisationer garanterat INTE gjort det)
...

Old manager:
assign work → coordinate people → review deliverables → decide

Human-agent leader:
define outcome → design roles and authority → orchestrate capacity → observe evidence → intervene in exceptions → improve the system

The key conceptual shift is subtle but important: the thing being managed is no longer primarily people or agents. It is the flow of agency through the organization.

That makes leadership less about controlling execution and more about designing the conditions under which execution can safely happen without you.

Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Canonical page

Verifiable MCP: an open invitation.

Joakim Jardenberg — Sat, 22 Aug 2026 12:00:00 GMT

Verifiable MCP: an open invitation.

https://sswcboken.se/

Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Canonical page

Local AI. Finally, a tool.

Joakim Jardenberg — Tue, 11 Aug 2026 12:00:00 GMT

Local AI. Finally, a tool.

2026-08-11 · Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

This turned out more LinkedIn than Facebook, so here you go ;)
"""
I’ve been tinkering with local AI models for three years now. And it has always ended the same way: either you have to assemble everything yourself like an IKEA cupboard with no instructions, or you get an answer that feels as though it came from an exhausted intern.

Now. Bionic from LM Studio + Muse Glimmer from Meta.

For the first time, I can actually imagine my mother-in-law running this at home without having to call me.

The interface is clean and fast, and you don’t need a PhD to work out where to click. And the model? It talks and thinks in a way that doesn’t make you dizzy with disappointment when you compare it with the big cloud offerings. It delivers.

Does it take a decent computer and plenty of memory? Yes. But the difference, the whole experience compared with what came before, is enormous. This no longer feels like a project. It feels like a tool.

Finally.

Original image · Swedish

Image description · AI: The original screenshot shows the writing prompt behind this post. JJ describes a usable local-AI interface and model, while acknowledging the hardware and memory requirements.

Original LinkedIn post

Original wording · sv

Det blev mer Linkedin än Facebook, så här får ni ;)
"""
Jag har petat med lokala AI-modeller i tre år nu. Och det har alltid slutat likadant: antingen får man bygga ihop allt själv som ett IKEA-skåp utan instruktioner, eller så får man ett svar som känns som att det kommer från en trött praktikant.

Nu. Bionic från LM Studio + Muse Glimmer från Meta.

Första gången jag faktiskt kan tänka mig att min svärmor skulle klara av att köra detta hemma utan att ringa mig.

Gränssnittet är rent, snabbt och man behöver inte ha en doktorsexamen för att förstå var man ska klicka. Och modellen? Den pratar och tänker på ett sätt som inte gör en yr av besvikelse när man jämför med de stora molngrejerna. Den levererar.

Krävs det en vettig dator och en hel del minne? Ja. Men skillnaden, helheten, jämfört med tidigare är enorm. Det här känns inte längre som ett projekt, det känns som ett verktyg.

Äntligen.

Joakim Jardenberg. English translation by AI. Original wording preserved. Titles and image descriptions added by AI.

Canonical page

Things I build. Questions I follow.

Joakim Jardenberg

I build things, work with people, and follow the questions that won’t leave me alone.

More than 90% of my work happens inside client organisations, helping people turn strategy, technology and AI into everyday practice. What I share here is a selection of my own explorations: tools, archives and experiments, built with AI.

Work with me Explore what I build

Now / 4 October 2026

From prompting to conversation

In Thursday's Interactions Matter keynote, I explored a progression: discovering agency, moving from prompts to conversation, learning to work as a team, and growing through honest feedback.

Read the latest note

Also in this note

What others are building: pocket agents

Follow the thought

In between the Sunday notes

Thinking out loud.

All notes

Short posts, useful discoveries and things I’m still figuring out.

09 OctLess managing the list. More doing.07 OctMarvin is paying attention.02 OctFriday admin has never been this much fun.

From ideas to everyday work

Strategy, implementation and adoption belong together.

I work with leaders and teams, and with the people who turn an idea into a way of working. What we learn there should change the next decision.

Find a way to work together

A few things I’m invested in

Two companies I care deeply about.

Over the past 30 years, I’ve made 100+ investments in early-stage startups. Today I’m down to two: Klang and Shiftic. Two that matter deeply to me, and two I use myself.

Klang

Investor and user. Klang is built by its founders and team.

Transcription, summaries and search that make conversations easier to work with. In my own work, I haven’t found anything that handles Swedish better.

What matters to me goes beyond the product. Smaller languages and regional voices deserve serious engineering too. Klang is putting that into practice with open model weights, open-source tools and a Swedish dialect dataset that others can use and build on.

Debattkollen shows what that can mean in public life: making political debates searchable, with transcripts and links back to the source. More people can find out what was actually said, inspect it and join the conversation.

“Joakim is one of my biggest inspirations. Ever curious about the world, he pushes you to the highest of standards”

Niklas SilfverströmCEO and co-founder, Klang

Explore Klang Open research Debattkollen

Shiftic

Investor and user. Shiftic is built by its founders and team.

A workspace for turning a change you want into behaviours people can practise, with plans, practical support and follow-up.

The gap between deciding that something should change and helping people actually do it runs through much of my work. Shiftic connects those intentions with the activities, materials and support people need.

If Shiftic takes hold inside organisations, I believe the human impact could be enormous: more people able to make worthwhile changes, with clearer expectations, better support and room to learn. Less energy lost between agreeing on a change and finding a way to make it happen.

“Joakim seems to have an incurable curiosity about technology, and it’s contagious. There is no one (honestly) who gets me to broaden my horizons and discover exciting new possibilities with technology quite like he does. He also challenges what needs challenging to make sure the business is ready for the future, creates value and has meaning. He does this even when it’s uncomfortable, but always with a great deal of care and generosity. I’m very grateful to have Jocke with us on our journey. Jocke, Mr Tough Love for entrepreneurs.”

Bella FunckCEO and co-founder, ShifticTranslated from Swedish

Explore Shiftic

Selected organisations

A few organisations I’ve worked with.

Different teams, different periods, different kinds of work.

Volvo Group
Mycronic
Atlas Copco
Storskogen
Novare
Teledyne FLIR
Lindinvent
Google
Lufthansa
Adidas
Pfizer
TUI
Volvo Cars
Boliden
Edukatus Alliance
Axel Johnson
Tieto
Hyper Island
Dole
Nowaste
Södra
Insight Events
Internetstiftelsen
Bron Innovation
Turistrådet Västsverige
Marknadsföreningen i Göteborg
Kraftkällan Evenemang
Cover and Move
Helsingborgs stad
Svalövs kommun
Burlövs kommun
Filipstads kommun

What others have said

A few things to explore

#ensakidag

An archive that keeps learning.

Det Nya Sverige

15 points for a better Sweden.

PPCP

Public context for people and AI.

Verifiable MCP

An experimental way for an AI client to check the origin and integrity of an MCP response.

Explore more projects

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

About JJ

Joakim Jardenberg

About

Still here. Still curious. More relevant than ever.

I’m Joakim Jardenberg. Most people call me JJ.

I’ve been building things with the internet since 1993. Today I work with leaders and teams around the world, helping them turn the possibilities of AI into something useful in their everyday work.

That takes me between strategy, hands-on implementation and the patient business of helping people change how they work. I’m an advisor, facilitator, speaker, builder and investor. What joins those roles is curiosity about what people could do next, and a willingness to get involved.

It started with making things

In 1993, I was handcrafting HTML and helping people find their way onto the web through Mackmeckarna in Malmö. We ran some of Sweden’s earliest web education, learning alongside librarians, architects and archaeologists. Each new website could give us a different idea of what this thing might become.

At Aftonbladet in the early web years, I worked with a small team to turn the paper’s text-TV news into an automatically updated web news flow.

For the 1994 Swedish election, my colleagues and I helped newspapers get connected to receive results by email. Much of my own work was at Aftonbladet, receiving and processing those results. Once people had the tools, they started finding other uses for them. I still recognise that moment today: someone tries a thing, sees what it makes possible, and begins asking a bigger question.

The early years, in my Internetmuseum interview

The world got bigger

That curiosity took me through Helsingborgs Dagblad, into running Mindpark, and later into public service as Helsingborg’s Head of Internet. It also took me around the world: speaking, teaching, advising and working alongside people in very different organisations.

Over the years, I’ve worked with companies including Google, Lufthansa, Volvo Group and Atlas Copco, as well as founders, smaller teams and public institutions. The setting changes, and so does what people need from me. Sometimes it’s a question that opens up a room. Sometimes it’s staying with a team long enough for a new way of working to take hold.

Some of the organisations and people I’ve worked with

The CEO whisperer

Some people call me a CEO whisperer. I like that description. A leader needs somewhere to think aloud, ask the unfinished question and have an assumption challenged. I bring experience, curiosity and a willingness to say what I see. We work out what matters and what to do about it.

Those conversations are most useful when I also spend time with the people expected to make change happen. I want to know what their work looks like, what helps and what gets in the way. What I learn with them belongs back in the leadership conversation. Strategy, implementation and adoption need to keep talking to each other.

Big change, many small moves

This is where much of my work lives today. We take something people actually need to do and try a different way of doing it. A colleague finds a useful application for AI, shares it, and someone else takes it further. We make time to practise, compare experiences and adjust. Small changes can travel a long way when people help each other.

Leadership matters here: giving direction, making room for experiments, removing obstacles and noticing what deserves to grow. I call my approach Guided Emergence, drawing on Henry Mintzberg’s thinking about how strategy is crafted through experience. We need enough direction to move together and enough openness to learn something we didn’t plan for.

That is why I work with sustained learning journeys, hands-on workshops and the everyday questions that follow them. I want people to leave more able to act, with the confidence and support to keep going. Their next attempt is part of the work too.

There’s a person in all of this

Family, equal human worth and the world younger people inherit matter deeply to me. So do play, generosity and a good conversation. I want technology to give more people the chance to understand, create and contribute. That matters in a global company, and it matters when one person discovers they can do something they thought was beyond them.

A fuller picture, in my public profile

And I’m still making things

These days, AI lets me turn more of that curiosity into things people can use. I build tools, open up archives and explore ideas in public. Much of my work happens inside client organisations; what you see on this site is a selection. Building keeps me close to the possibilities, the awkward details and the surprises.

Back in 1993, I made websites by writing HTML by hand. This one has grown through conversation with AI. After all these years, I can still sit here and be amazed by what we can do.

I have never worked as much as I do today, and I have never had this much fun. What a time to be alive.

In The next 5,000 days of the web, Kevin Kelly looked at the abundance the web had brought us and observed that it “is amazing, and we're not amazed.”

I hate to argue with Master Kevin Kelly but he was wrong. I am amazed.

Let’s work on something together

Elsewhere online

The conversation continues.

LinkedIn Facebook Instagram X / Twitter YouTube Mastodon

A few things I’m invested in

Two companies I care deeply about.

Over the past 30 years, I’ve made 100+ investments in early-stage startups. Today I’m down to two: Klang and Shiftic. Two that matter deeply to me, and two I use myself.

Klang

Investor and user. Klang is built by its founders and team.

Transcription, summaries and search that make conversations easier to work with. In my own work, I haven’t found anything that handles Swedish better.

“Joakim is one of my biggest inspirations. Ever curious about the world, he pushes you to the highest of standards”

Niklas SilfverströmCEO and co-founder, Klang

Explore Klang Open research Debattkollen

Shiftic

Investor and user. Shiftic is built by its founders and team.

A workspace for turning a change you want into behaviours people can practise, with plans, practical support and follow-up.

“Joakim seems to have an incurable curiosity about technology, and it’s contagious. There is no one (honestly) who gets me to broaden my horizons and discover exciting new possibilities with technology quite like he does. He also challenges what needs challenging to make sure the business is ready for the future, creates value and has meaning. He does this even when it’s uncomfortable, but always with a great deal of care and generosity. I’m very grateful to have Jocke with us on our journey. Jocke, Mr Tough Love for entrepreneurs.”

Bella FunckCEO and co-founder, ShifticTranslated from Swedish

Explore Shiftic

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Changes

Joakim Jardenberg

Changes

A site that keeps moving.

0.5.0

27 September 2026 · Editorial site renewal

Dark editorial design, with a responsive sidebar and accessible navigation.
A public-project catalogue, practical idea pages and site search.
A dated Now notebook, with source links and clear editorial status.
Public profile and MCP onboarding, with copyable connection details.
Direct routes into working together, source credits and clearer authorship.

The renewed site brings the public work, ideas, investments and weekly notebook together. Photographs remain provisional while a new set is being prepared.

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

How this was made

Joakim Jardenberg

How this was made

AI + JJ.

AI does substantial work here. I help shape what we make, question it and decide what deserves to go out into the world.

This site grew from a review of the work, the public projects and the thinking behind them. The research, writing, design exploration and implementation happen in our AI collaboration. GitHub holds the source; Lovable provides the site’s platform.

The photographs are real. Project images come from their original public sites. New editorial explanations are credited as AI + JJ. Original human work and other people’s ideas keep their own attribution.

A site that can change its mind

The Now notebook is drawn from recent work and conversations. It is edited for public use, dated, and kept separate from private source material. A developing thought should remain a developing thought.

The project catalogue and idea pages have a maintained content structure. They can change as the work changes, without rebuilding the site around a new set of slogans.

Sources and corrections

My public profile is the fuller, sourced account of my background and preferences. Intellectual origins and collaborator credits appear beside the relevant work.

If something is wrong or missing, tell me what you found. An accurate account matters more than a tidy one.

Images

Portrait photography: Kristofer Samuelsson. Project imagery is identified with the linked project and its original publisher. It illustrates the project; it is not a claim of exclusive authorship.

Open by design

My own writing is licensed CC BY 4.0. Keep the credit, link the source and identify your changes. Other people’s quotations, photographs and media keep their own rights.

Feeds, downloads, accessibility and reuse

Version

0.5.0 · Read the changes

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

For your AI

Joakim Jardenberg

For your AI

Bring some context into the conversation.

Your AI can read my public profile: how I work, what matters to me, and the sources behind it. Use the interface that suits you.

Start with something useful

Before we work together, ask your AI what might help us collaborate well. Or use it to explore an idea you found here. You don’t need to read a protocol specification first.

This site, in your AI

The same public MCP also provides the published pages on this site, including every Now edition and Note, with original languages and image descriptions. Ask it to list the collection, search a question, then read the relevant page. Sources, dates and AI credits travel with the content.

Feeds, downloads and reuse

Connect the public MCP

If your AI application supports remote MCP servers, add the endpoint below through its connector or MCP settings. Exact steps and availability differ between applications.

Open the setup guide and choose the instructions for your application.
Add the endpoint. This is a public, read-only source of context.
Ask the first question above, starting with the short version and relevant sections.
Check the answer’s source links and ask a follow-up.

Open the current setup guide

What you are connecting to

The server provides the profile, individual sections, sources and version information, plus search and reading tools for the published jardenberg.com collection. It gives access to the public material I have selected, not my private workspace or conversations. It cannot act or make commitments on my behalf.

Some clients have access limitations. The setup guide records those limitations and links the current official client instructions. If a connection fails, the readable profile and downloads remain useful alternatives.

The idea behind it

PPCP is a draft Public Personal Context Protocol. It explores how people can carry a sourced, maintainable public account of themselves into different AI systems. The draft protocol and my own profile are separate parts of that work.

Explore the project

Let services meet a person in the interface they choose, with useful context and control intact.

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Four Guiding Principles

Joakim Jardenberg

All ideas

Ideas / In practice

Four Guiding Principles

A memorable starting point for bringing AI into everyday work while keeping human judgment in view.

These principles give people a shared starting point. Their usefulness comes from trying them on real work and discussing the tensions they reveal.

The foundation comes from Ethan Mollick's Co-Intelligence. The teaching wording and examples here are my adaptation, developed through practice.

In September 2026 I added a refinement to the third principle: how you address AI shapes the interaction. Context, expectations and language influence what happens; this does not mean every conversation is used to train the underlying model.

I am also exploring what being the human in the loop requires when AI can act: where judgment belongs, how authority is bounded and how we learn from consequences. That is work in progress, alongside the established four principles.

Put it to work

Invite AI into everything you do

Explore where it could help with thinking, making, learning and everyday tasks. Invitation still calls for judgment about the task and its context.

Be the human in the loop

Keep responsibility for how AI is used, the judgments that matter and the process around delegated work.

Treat AI like a human, but never forget it is not

Give context and feedback as you would to a colleague, while keeping its limitations in view. How you address it shapes the interaction.

The AI you have today is the worst AI you will ever have

A reason to keep learning and revisit what is possible, not a promise that every release or answer will improve.

Where this comes from

Adapted by JJ from Ethan Mollick's Co-Intelligence. The September refinement also draws on Nicklas Berild Lundblad's tulpa essay. Co-existence extensions remain working proposals.

Editorial explanation · AI + JJ

Explore this in your own work More ideas

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

The Four Jobs

Joakim Jardenberg

All ideas

Ideas / In practice

The Four Jobs

Do the work. Improve the work. Grow yourself. Develop others.

Most of us can describe the work waiting for us today. It is often harder to protect time for changing how that work happens, developing our own capability or helping someone else grow.

The Four Jobs makes those different kinds of contribution visible. AI can have a role in all four. A useful question is where it creates room for better work, and what we choose to do with that room.

Try mapping a real week. Where did the attention go? Which job keeps getting postponed? The point is to examine your own situation, not to measure everyone against an invented ideal percentage.

Put it to work

Do the work

Deliver the task, make the decision and meet the commitment in front of you.

Improve the work

Change the process, remove friction or build something that makes the next attempt better.

Grow yourself

Develop the knowledge, judgment and capability you will need next.

Develop others

Help colleagues learn, share useful practice and strengthen the team's ability to act.

Where this comes from

JJ's working framework for leadership, learning and AI adoption. This page uses a self-assessment question rather than an unsupported population statistic.

JJ's public profile

Editorial explanation · AI + JJ

Explore this in your own work More ideas

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Guided Emergence

Joakim Jardenberg

All ideas

Ideas / In practice

Guided Emergence

Set direction, make room for people to try things, and let what they learn change the next decision.

I find strategy most useful when it stays in contact with the people doing the work. They encounter the awkward details, find the useful shortcuts, and discover possibilities that were difficult to see from a planning meeting.

Guided Emergence is how I try to keep that contact alive. Set a direction and make the important boundaries clear. Give people room, authority and resources to explore. Pay attention to what happens. Then use what you learn to decide what deserves more support, what needs changing and what should stop.

Strategy, implementation and adoption belong in that same conversation. Each gives the others something to work with.

Imagine a team preparing for customer conversations. Information is scattered, and someone suggests an AI assistant. First agree what better preparation would mean: useful questions, fewer missed commitments, less hunting for information. Decide what the assistant can access and how the team will notice a poor result.

Give a few people time to try it on real work. Perhaps it produces a useful briefing. Perhaps it reveals that nobody agrees which records are current. Either finding can change the next move: extend the experiment, improve the information or stop. This is a hypothetical example, not a claim about a client result.

The practical test is whether learning can influence decisions. There should be a route from an experiment to a change in direction, resources or working practice.

Put it to work

Guiding principles

Establish direction and boundaries that help people make decisions without scripting every action.

Experimentation

Make room to try something on real work and learn from what happens.

Decentralized decisions

Give people meaningful authority close to the work.

Feedback and learning

Listen to colleagues, customers and users, then let their experience change the next move.

Resources for exploration

Reserve actual time and support for exploration.

Scaling what works

Look for evidence before helping more people adopt a promising practice.

Adaptive strategy

Revisit direction, commitments and priorities as understanding improves.

Cross-functional collaboration

Bring together people who see different parts of the problem.

Enabling technology

Use tools to make the work, evidence and collaboration easier.

Where this comes from

JJ's formulation, developed in 2023 and informed by Henry Mintzberg's Crafting Strategy. This account does not claim exclusive coinage of the term.

Editorial explanation · AI + JJ

Explore this in your own work More ideas

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Interface Sovereignty

Joakim Jardenberg

All ideas

Ideas / Working idea

Interface Sovereignty

Let services meet a person in the interface they choose, with useful context and control intact.

I want my sources, services and tools to come to an interface that fits how I think and work. Moving between products should not mean starting the explanation of my life and my task from zero every time.

That can involve text, voice, images, files, APIs or MCP. The important question is whether the person gains continuity and practical agency.

There are real design questions here. A convenient interface can hide disagreement between sources, conceal the context of an action or make it harder to see what actually happened.

This is a working thesis and a research question. A conversational interface that suits me is not automatically the right interface for everyone.

Put it to work

Keep context useful

Preserve continuity across tasks and devices without making the person repeat everything.

Keep sources reachable

Let people inspect the original material and notice when accounts disagree.

Keep actions legible

Distinguish suggestions, authorization, attempts and completed actions.

Keep choice real

Make it possible to correct, replace or leave the interpreting layer.

Where this comes from

JJ's chosen-interface thesis, expressed in 2026. The design questions are an AI + JJ working interpretation, not a claim of legal sovereignty or universal accessibility.

JJ's public profile: chosen interface and memory

Editorial explanation · AI + JJ

Explore this in your own work More ideas

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Six Value Vectors

Joakim Jardenberg

All ideas

Ideas / In practice

Six Value Vectors

Look for AI's value in decisions, energy, memory, risk, development and access, as well as time saved.

Saving time can matter. It also leaves some important questions unanswered: did the decision improve, did a person gain confidence, or could someone finally participate in work that had been difficult to access?

The Six Value Vectors give a team several places to look for value. They help make a conversation about AI more specific: what should change here, for whom, and how would we know?

Choose evidence that fits the work. A faster decision is not automatically a better decision; a lower workload is not useful if the quality disappears. The vectors are lenses for inquiry, not a guarantee of return.

Earlier material calls this the Five Value Vectors. Access and Inclusion was added in July 2026. Five Ways AI Changes Work is a different framework.

Put it to work

Decision velocity and quality

Help people reach timely decisions with better information and clearer reasoning.

Cognitive capacity and wellbeing

Reduce avoidable mental effort and make room for work that needs attention.

Organizational memory and knowledge activation

Make accumulated knowledge easier to find, connect and use.

Proactive compliance and risk mitigation

Notice relevant obligations, inconsistencies and emerging risks earlier.

Talent lifecycle transformation

Support people as they learn, develop, contribute and share what they know.

Access and inclusion

Change who gets to participate and act, including across language, confidence and capability barriers.

Where this comes from

JJ's value framework, extended from five to six vectors in July 2026. Explanations here are AI + JJ editorial adaptations.

JJ's public profile: what counts as value

Editorial explanation · AI + JJ

Explore this in your own work More ideas

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Stacked Principles

Joakim Jardenberg

All ideas

Ideas / In practice

Stacked Principles

Several commitments can be genuine. The decision is which one takes precedence here, and why.

I can value the quality of a service and also care about what my spending helps enable. Usually one principle guides the choice. At a particular boundary, another becomes more important.

Stacked Principles gives me a way to explain that change in priority. It is contextual judgment, with reasons attached. The other commitment has not stopped mattering.

The stack is neither a permanent numerical ranking nor a flat list of values. Scale, consequences, available alternatives and the connection between an action and its effects all matter.

My choice for myself is also different from asking everybody else to make the same choice. A public campaign has different stakes and needs a different evidential basis.

Put it to work

Name the ordinary principle

What would normally guide this decision?

Identify the boundary

What consequence or new evidence makes another commitment decisive?

Give the reason

Why does that priority make sense here, considering scale and alternatives?

Keep the decision revisable

What could change the conclusion, and would you reason similarly in a comparable case?

Where this comes from

Grounded in JJ's own reasoning and an AI-assisted explanation he endorsed. The practical questions are an editorial operationalization, not verbatim quotations or a fixed constitution.

Stacked Principles in JJ's public profile

Editorial explanation · AI + JJ

Explore this in your own work More ideas

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Notes

Joakim Jardenberg

Notes / From LinkedIn

Thinking out loud.

Short posts, useful discoveries and things I’m still figuring out. A little closer to the moment.

English translations by AI. Original posts and images are preserved. For the slower reflection, read Now

Subscribe to Notes by RSS All feeds & downloads

14 notes · 11 August 2026–9 October 2026

The latest

9 October 2026

Less managing the list. More doing.

AI translation · Swedish → English

Original image · EnglishClick to enlarge

Open this note Läs originalet LinkedIn

7 October 2026Everyday AI

Marvin is paying attention.

AI translation · Swedish → English

Read note Läs originalet LinkedIn

2 October 2026Everyday AI

Friday admin has never been this much fun.

AI translation · Swedish → English

Folks. What a time to be alive. Friday admin has never been this much fun.

Read note Läs originalet LinkedIn

28 September 2026Building & openness

A little hack to follow all the other hacks.

AI translation · Swedish → English

Read note Läs originalet LinkedIn

25 September 2026Learning & leadership

A “D-level decision maker”. Fair enough.

Original in English

One of my AIs recently called me a ”D-level decision maker”. I’d been telling it off for failing to prepare my day properly. It had a few things to say about my contribution to that failure. Fair enough. Let’s put it on the table. What followed was a…

Read note LinkedIn

23 September 2026Building & openness

Klang. Those geniuses.

AI translation · Swedish → English

Read note Läs originalet LinkedIn

22 September 2026Learning & leadership

Get AI under your skin.

AI translation · Swedish → English

It’s a bit embarrassing to throw leadership quotes around, but this is too important to ignore. Myron Tribus said: “You cannot lead what you do not understand, and you do not understand what you haven't done.” I think that explains a large part of the partly…

Read note Läs originalet LinkedIn

31 August 2026Learning & leadership

Learning, at any hour.

Original in English

Read note LinkedIn

24 August 2026Small things

Some things are small enough to fix.

AI translation · Swedish → English

Read note Läs originalet LinkedIn

22 August 2026Building & openness

Verifiable MCP: an open invitation.

Original in English

Read note LinkedIn

22 August 2026Learning & leadership

The flow of agency.

AI translation · Swedish passages

Read note Läs originalet LinkedIn

14 August 2026Everyday AI

A new Chief of Staff.

AI translation · Swedish → English

I’ve hired a new “Chief of Staff”. Always be learning…

Read note Läs originalet LinkedIn

13 August 2026Building & openness

The dreamers of Tjärö, sixteen years later.

AI translation · Swedish → English

Read note Läs originalet LinkedIn

11 August 2026Everyday AI

Local AI. Finally, a tool.

AI translation · Swedish → English

This turned out more LinkedIn than Facebook, so here you go ;) """ I’ve been tinkering with local AI models for three years now. And it has always ended the same way: either you have to assemble everything yourself like an IKEA cupboard with no instructions,…

Read note Läs originalet LinkedIn

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Now

Joakim Jardenberg

Now

What I’m actually thinking about.

A developing question, something made or learned, and room for what I haven’t figured out yet.

Subscribe to Now by RSS All feeds & downloads

4 October 2026Weekly noteAI + JJ

From prompting to conversation

In Thursday's Interactions Matter keynote, I explored a progression: discovering agency, moving from prompts to conversation, learning to work as a team, and growing through honest feedback.

What others are building: pocket agents

Microsoft's Autopilot announcement points in the same direction. I will be very sad if it doesn't bring back Clippy.

A list of links is a beginning

A signature isn't the whole answer

AI + JJ. A dated editorial account of the work and thinking, with sources linked in the text.

The notebook

4 October 2026From prompting to conversation 27 September 2026Where does human judgment belong?

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Open, accessible and yours to explore

Joakim Jardenberg

Read it your way

Open. And meant to be used.

Follow the writing in your own reader. Bring it into your AI. Download it, quote it, build on it.

Subscribe without an account

These are full-text feeds, with original dates, source links, images and AI credits. New published material appears when the site is updated. Choose the pace that suits you.

Now + Notes: all the writing
Now: the weekly reflection
Notes: thinking closer to the moment
Everything: the whole public site, including ideas, projects and reference pages
JSON Feed, if your reader prefers it

Take a copy

Download the public collection as JSON or read it all as Markdown. The compact index links to a separate Markdown copy of every page. Notes include the original wording as well as the labelled English translation.

The exports are generated from the same published content as this site. They include public pages only. Linked archives and projects have their own collections; their contents are not silently copied here.

Bring your AI

The public, read-only MCP is at joakim.jardenberg.net/mcp. It provides my public profile and tools to list, search and read this site’s published material. Start with a search or a relevant page, then follow the sources.

Setup and a first question. Signed responses preserve the content, source, rights and AI credit. A signature establishes where a response came from and whether it changed, not whether every claim is true.

Reuse with credit

My own writing on this site is licensed Creative Commons Attribution 4.0 International (CC BY 4.0). You may copy, adapt and reuse it, including commercially. Credit Joakim Jardenberg, link the original and the licence, and identify changes. Preserve “AI + JJ” or the translation credit where it applies.

Other people’s quotations, recommendations, photographs, screenshots, logos and other third-party material retain their own rights. Their appearance here does not put them under this licence or imply endorsement of your reuse. Portrait photography is by Kristofer Samuelsson. Credits and intellectual origins stay beside the work.

Access is part of the work

The site supports keyboard navigation, a skip link, system light and dark appearance, reduced motion, readable page structure and text descriptions for images. The public reading pages and Swedish originals are available without JavaScript; search, filters, copy buttons and the booking embed use it. Direct booking links remain available.

Accessibility needs continued testing with people and assistive technology. If something gets in your way, tell me where and what happened. Alternative reading formats are linked above.

Made to be found

robots.txt allows all crawlers across the public site. The sitemap lists every reading page, including Swedish originals. Feeds are advertised in each page’s metadata. Host discovery links the public profile, MCP server card, feeds and collection. The MCP discovery file points to the existing server and its authoritative signing-key discovery.

Public reading pages permit full search snippets and large image previews. Search results and missing pages are excluded from indexing. Discovery conventions such as PPCP and llms.txt help supporting clients; they cannot guarantee that every AI system will find or index the site.

A known limit

On 9 October 2026, the hosting service blocked requests from the default Python urllib client. A descriptive User-Agent or an empty User-Agent worked in the checks. No login or API key is required. This hosting restriction is unresolved; access for every automated client is not yet guaranteed.

How this was made · What changed

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Things I’ve built

Joakim Jardenberg

Things I’ve built

A useful thing. An open question. Sometimes both.

More than 90% of my work happens inside client organisations. The projects here are my own explorations: tools, archives, public ideas and experiments, built with AI and shared for others to explore.

A map and companion for a particular food walk through Helsingborg.

Public ideas · Live

Future of Trucking · Job Fit Atlas

An exploratory field guide to changing truck-driver preferences and possible job fit in the United States.

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

2026GPT

Joakim Jardenberg

All projects

Experiments / Experimental

2026GPT

A proof of concept for what an internal AI assistant could feel like when the product experience comes first.

This experiment brings together chat, documents, tools and model routing around a fictional industrial company. It asks how much better the everyday experience of an internal AI system could be.

The project builds on LibreChat and other open-source components. The public repository is the place to inspect the experiment; this listing makes no promise of a currently available hosted demo.

Explore further

LibreChat upstream

These are proposals I am putting into the conversation. They should be read as a civic contribution, with room for disagreement and further work.

Make the thinking available

Publishing the proposals together makes their connections visible. It also gives a reader something specific to respond to: a proposition, its reasoning and the questions it raises.

The site is part of the work. Making an idea easier to inspect is a useful step towards finding out whether it holds up.

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

#ensakidag

Joakim Jardenberg

All projects

Archives / Live

#ensakidag

Earlier thinking, new connections, and an archive you can explore with AI.

I started #ensakidag in 2017 because I needed to get out of my own head for a while. I had been deep in a large project in Palma, and talking to a camera became a way to think aloud with other people.

The format was simple: take one thing happening in technology or on the internet, explain why it matters, and ask what it might mean for us. In Swedish, usually in a few minutes. Occasionally I get carried away.

Those reflections became a record of what I was noticing, wondering about and sometimes getting wrong. When I returned to the project in 2026, I wanted to work with that history: find a thought again, follow a connection, see what had changed.

The archive started with a practical need

I wanted my AI assistants to have access to my earlier material. That was the starting point for the archive and its MCP connection. Then it made sense to give people a website through which to explore it too.

You can search the archive, read episode transcripts and follow related subjects. A compatible AI assistant can also retrieve the material. Original episodes sit alongside AI-produced summaries, connections and retrospective commentary.

A record of thinking, with its dates attached

Something I said years ago is a record of what I said then. A later AI assessment is another layer, with its own limitations. Both can be interesting without pretending they are the same thing.

For me, part of the pleasure is discovering something I had forgotten saying. Another part is seeing what someone else makes of it. A conversation can keep developing long after the recording ends.

A useful place to start

Episode 500 tells the story behind making the archive useful again. Or bring your own subject: find an earlier episode, look at the original material and ask which question is still worth exploring.

Explore further

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Future of Trucking · Job Fit Atlas

Joakim Jardenberg

All projects

Public ideas / Live

Future of Trucking · Job Fit Atlas

An exploratory field guide to changing truck-driver preferences and possible job fit in the United States.

The guide explores changes in drivers' preferences and behaviour, and how different kinds of work might fit those preferences in the years ahead.

Use it to develop questions and examine possibilities. It is a general public research project, not a hiring assessment or an employer's promise about future roles.

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Gör mig skånsk

Joakim Jardenberg

All projects

Experiments / Live

Gör mig skånsk

A playful experiment in language, voice and sounding a little more Skåne.

An idea about regional language became something people could try. Gör mig skånsk explores what AI can do with a distinctly local voice.

It is an invitation to play with language. A generated result is an interpretation of a dialect, with all the variation and imperfection that implies.

Explore further

DialectLab source

Explore further

Upstream integration

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

PalmaStories

Joakim Jardenberg

All projects

Public ideas / Live

PalmaStories

A slow journal of Palma and Mallorca, with places, people and details worth spending time with.

A place can become familiar through the things you notice together: a shop, a meal, a street, someone who makes something well. PalmaStories gives those observations room to become stories.

The writing keeps its bylines. Kattis's stories and perspective belong to her; my part in the site is helping make a home for the work.

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

PPCP

Joakim Jardenberg

All projects

Infrastructure / Experimental

PPCP

A draft protocol for sharing public personal context with people and AI.

Every new AI conversation can begin with the same small problem: it knows a great deal about the world and very little about the person in front of it.

Public Personal Context Protocol is an open draft exploring how a person can publish context that is useful to both people and AI systems. The specification, schemas and examples give others something concrete to inspect and try.

My public profile is a reference implementation. It brings together how I work, what matters to me and how I want to be challenged, with readable pages, downloads and read-only MCP access.

Try it with a real question

Connect the public profile to a compatible AI assistant, then ask how to prepare for working with me. A useful answer should point back to the published material and keep my stated preferences distinct from the assistant's interpretation.

You can read the same context without connecting anything. The website and downloadable versions remain part of the project.

Keep ownership and interpretation visible

An approved profile is my account of myself. It is not independent proof of every claim, and an AI's later adaptation does not automatically become my new position.

PPCP remains an open draft. The point of making it public is to let people examine the proposal and learn from using it.

Explore further

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Prompt Pantry

Joakim Jardenberg

All projects

Tools / Live

Prompt Pantry

Build prompts from reusable ingredients, then adapt them to the thing you actually want to do.

I like the pantry metaphor because the contents are there to be used. Prompt Pantry brings that idea to working with AI: gather useful ingredients, combine them, try the result and refine it.

The aim is to help you make a prompt that fits your work, while learning something about what makes the interaction useful.

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

My public context

Joakim Jardenberg

All projects

Infrastructure / Live

My public context

A fuller account of who I am, how I work and what helps a collaboration with me.

A biography can tell you where someone has worked. This profile also tries to explain how I think, what matters to me, and what a useful working relationship looks like.

The public edition is available to read, download or connect to an AI assistant. I have reviewed the published account; its editorial preparation and source distinctions remain visible.

Explore further

MCP setup and first questions

Explore further

JJ's original contribution

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Valvakan 2026 · Partivakor

Joakim Jardenberg

All projects

Archives / Live

Valvakan 2026 · Partivakor

A public research collection about the parties' election-night gatherings in 2026.

A focused collection makes material from the 2026 election-night context easier to inspect together. It is a dated research project with its own scope.

Read entries in the context of their sources and dates. This is not a live election-results service.

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Verifiable MCP

Joakim Jardenberg

All projects

Infrastructure / Experimental

Verifiable MCP

An experimental way for an AI client to check the origin and integrity of an MCP response.

When an AI retrieves something from an archive, I want the connection to the source to remain inspectable. Which server supplied it? Has the response changed? Does it contain someone's original words, or an AI's later interpretation?

Verifiable MCP is a working pattern for attaching a digital signature and provenance to an MCP response. The signed material includes declarations about the source, rights and freshness, so those declarations travel with the content being checked.

A valid signature can establish that the response matches what the signing server issued. The reader still needs to evaluate the source and the claim. Signing a statement does not make it true.

Keep the source and its labels together

The distinction between an original contribution and an AI-written response matters. So do the rights attached to each. The pattern places those labels inside the signed material, where changing them can be detected.

For my own archive, a check can establish what my server supplied. For an index of other people's publications, it establishes what the index supplied. It does not prove that the original publisher wrote or approved the index's representation.

Tried with different kinds of public knowledge

The public repository documents reference implementations for #ensakidag, the SSWC book and the RISE/AI Sweden connector. They bring different situations: my own material, contributions by many authors, and an index of third-party publications.

That variation is useful. A common verification pattern needs to preserve the differences between sources, authorship and rights. Each server's current implementation and scope still need their own check.

Something you can inspect and test

The repository contains the specification, Python and JavaScript verifiers, a conformance checklist and published test vectors. The tests include altered and invalid responses that a verifier must reject.

Start with the explanation, run the published examples, then examine a reference server or try the pattern with a corpus you operate. The useful feedback is what the test reveals, including where the implementation falls short.

An experimental pattern with a defined scope

The signature covers the server response. It does not certify the answer an AI later writes from it. A client also needs to perform verification; simply receiving signed metadata is not the same as checking it.

The specification remains experimental. It describes its trust assumptions and limits, and makes conformance something to test rather than a badge to assume.

Explore further

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page

Work with me

Joakim Jardenberg

Work with me

Bring the question you’re trying to answer.

We can start with a conversation, a room full of people, or a real piece of work. The useful format follows from what needs to change.

Send me a note Book a conversation

Let’s find a time

Book a conversation.

Choose 25 or 50 minutes, then pick a time. Tell me what you’d like us to work on when you book.

The calendar opens here when you choose a meeting length.

Prefer a separate window, or having trouble?

25 minutes on Cal.com (opens a new tab)50 minutes on Cal.com (opens a new tab)

Keynotes & live conversations

Conferences, leadership gatherings and teams opening a new conversation.

A shared starting point for a room that needs to think clearly about what AI changes.

We begin with the audience and the question the event needs to address. I shape the examples and argument around that context.

The format can include a keynote, demonstration, conversation or a combination. The useful question is what people should be able to understand or try afterwards.

Workshops & Intellectual Boot Camps

Teams ready to move from discussing AI to working with it together.

Bring actual work into the room and explore what becomes possible with AI.

Participants work with real questions, tasks and material. We make room for both practical exploration and discussion of what the experience means.

For leadership teams, an Intellectual Boot Camp gives more concentrated time to work through strategic, organizational and operational questions. Scope and follow-up are designed around the team.

Learning Journeys & Copilot Convoys

Leaders and teams who want adoption to develop over time, with support close to the work.

Sustained practice, shared experiments and a rhythm for learning from real work.

Regular sessions and smaller experiments give a group time to build capability, bring difficulties back and learn from one another.

The design can combine strategic conversations, a Convoy, informal AI-Fika and support between sessions. The rhythm should fit the work and help the organisation carry the learning forward.

Strategy & advisory work

Leadership teams making consequential choices about AI, working practices and organizational capability.

Connect leadership choices with what implementation and frontline experience are revealing.

We examine direction, priorities, assumptions and the conditions people need to act. That can lead to an AI Compass, guiding principles, a decision or a carefully bounded experiment.

Guided Emergence keeps strategy connected to practical learning. A useful deliverable should help the organisation make its next decision and recognize what would change it.

Facilitation & moderation

Workshops, panels, management conversations and gatherings that need active participation.

Help a group get into the conversation it needs, with enough structure to move somewhere useful.

The work begins with the purpose of the gathering and the people in it. Different perspectives, uncertainty and disagreement can be useful material.

We shape the format around the conversation, then make the decisions, learning and open questions clear enough to carry forward.

Build something useful

A clear problem worth exploring through a small, usable build.

Take a concrete problem far enough that we can try a working response to it.

A prototype, a focused tool or a better route into existing knowledge can help make a question tangible. We begin with what someone needs to do.

AI does much of the making. We shape the purpose, inspect the result and learn from using it. The next step depends on what that experience shows.

Selected organisations

A few organisations I’ve worked with.

Different teams, different periods, different kinds of work.

Volvo Group
Mycronic
Atlas Copco
Storskogen
Novare
Teledyne FLIR
Lindinvent
Google
Lufthansa
Adidas
Pfizer
TUI
Volvo Cars
Boliden
Edukatus Alliance
Axel Johnson
Tieto
Hyper Island
Dole
Nowaste
Södra
Insight Events
Internetstiftelsen
Bron Innovation
Turistrådet Västsverige
Marknadsföreningen i Göteborg
Kraftkällan Evenemang
Cover and Move
Helsingborgs stad
Svalövs kommun
Burlövs kommun
Filipstads kommun

For an engagement like yours, ask me for relevant references and examples we can share.

In their words

What it’s like to work with me.

Selected recommendations, shared directly and on LinkedIn.

“Joakim seems to have an incurable curiosity about technology, and it’s contagious. There is no one (honestly) who gets me to broaden my horizons and discover exciting new possibilities with technology quite like he does. He also challenges what needs challenging to make sure the business is ready for the future, creates value and has meaning. He does this even when it’s uncomfortable, but always with a great deal of care and generosity. I’m very grateful to have Jocke with us on our journey. Jocke, Mr Tough Love for entrepreneurs.”

Bella FunckCEO and co-founder, ShifticTranslated from Swedish

“Joakim is one of my biggest inspirations. Ever curious about the world, he pushes you to the highest of standards”

Niklas SilfverströmCEO and co-founder, Klang

“Joakim is an outstanding moderator with a remarkable ability to be both professional and personal in his delivery.”

Stefan BromanInsight EventsTranslated from Swedish

“The best and most engaged speaker we have contracted.”

Eva SundqvistTieto

“He is not just talking about change, he embodies it on a daily basis.”

Jakob WiderbergVolvo Cars

“Joakim is a devoted Internet evangelist, always looking for new ways to stretch our minds and try things in a different way.”

Anne-Marie Eklund LöwinderIVA

“He has helped us to realize much about our business and given us the advice, ideas and criticism to help improve what we do. I fully recommend him.”

Nicklas Berild LundbladGoogle

“If you need a creative thinker and a person that actually delivers his visions, retain Joakim.”

Johan HübnerDelphi

Before we meet

Send the question, who is involved, the relevant context and what you hope will be different afterwards. Unfinished thinking is welcome. English and Swedish both work.

I usually prefer an asynchronous start and a booked conversation. It gives us both a chance to arrive prepared.

More about working with me

AI + JJ. Original human work and other people’s ideas retain their attribution.

Canonical page