We may be entering an age in which publishing also means making what you know available as pieces that people, and machines, can assemble around a need you never anticipated.
Build knowledge resources, tools and apps from what you’ve already written, keeping your sources, reasoning and checks visible.
I help experienced independent consultants, advisors and small expert-led firms that already deliver valuable work, but whose online presence fails to show why their expertise is different and worth paying for.
In Short
I think articles will gradually become the way we introduce and explain a larger resource we are building. A collection of useful tools. A directory of people we recommend. A map of a subject. A guide that helps someone make a decision.
This is where I see an opportunity in liquid content: knowledge organized so that it can be used in different ways, according to what someone needs.
But when you take knowledge out of an article and put it into another form, you can lose who said it, how they reached that conclusion, and whether anyone checked it. A recommendation can keep circulating long after the reasons for trusting it have disappeared.
For an independent expert, I think the opportunity is to build useful resources that preserve those reasons: the original sources, your process, what you verified and when. The article helps someone discover the resource. The resource helps him do something, and gives him enough information to judge whether he can trust it.
Table of Contents
1. What Is Liquid Content?
Liquid content is information that can be put together in different ways depending on the person, the question or the situation.
The term has been used with different meanings. Giorgio Taverniti’s Google liquido looks at Google as a whole ecosystem, where the same subject can be discovered through different services and in different forms.
In publishing, the idea goes further. The Reuters Institute, in its 2026 report on journalism and technology, describes a change:
“away from authoring ‘articles’ towards more flexible atomic objects.”
Think of the separate things inside an article: a finding, an explanation, an example, a recommendation, the evidence that supports it. These are the knowledge objects that can be connected and used again.
Turning an article into an audio recording gives someone another way to read/hear the same material. Building a decision guide from the recommendations in ten articles gives someone a different way to use what you know.
AI can help assemble those pieces into an answer to a question you never wrote an article about. But the sources and the conditions behind each answer need to remain available too.
2. What Changes for an Independent Expert
Most of the discussion I found about liquid content concerns news publishers. How can they adapt a story for different readers? How can they offer a shorter explanation, an audio version, or more detail about one part?
I want to look at what this means if you are an expert, a consultant, an advisor or a journalist working on your own.
You probably already have useful material spread across your articles. Recommendations you have explained. Sources you trust. Cases you have studied. Questions you have answered more than once.
But someone coming to your work for the first time has to find all that, read it, and put it together himself. He may not even know which articles contain what he needs.
I think that is a job worth doing for him.
If you compare tools, the comparisons could become a resource where he can choose according to his needs. If you follow a field closely, your research could become a map showing the people involved, the ideas they are developing, and where they disagree.
If you are a journalist following one story, you could maintain a timeline with the original documents, the claims that have been checked, and the questions that remain unanswered. Each new report would add to something a reader can consult later.
My claim is that these resources will gradually take over more of the practical work we now expect articles to do. The article will often introduce the subject, explain a change, show one useful example, and help someone discover the resource. It can also help sell access to it.
I would still want the article to be useful on its own. But I would give the fuller comparison, the organized evidence and the decision guide a place where I can keep improving them.
3. Three Examples You Can Look At
a) Tali Režun: Answers Collected Across His Own Writing
Dr. Tali Režun’s Field Notes organizes material from his published work into nine chapters about building and working with AI systems.
Each chapter starts with a question and an answer. The sources are alongside it, so you can go back to the writing it came from. You can also download the material and ask questions of it in your own AI system.
He explains the relationship to his articles plainly:
“It is distilled across them, and it links back.”
You can start with your question, without having to read his articles in the order he published them.
b) Sangeet Paul Choudary: Several Ways To Explore One Book
On May 31, Sangeet Paul Choudary published an explanation of his interactive companion to Reshuffle.
He calls the work behind it idea architecture: organizing ideas so that they can be combined in different ways for different purposes.
His companion site includes a map of the concepts and pathways through the material. The book gives you a sequence to read. These other forms let you explore according to what you need to understand.
I had already recommended his thinking in my June 3 issue. Liquid content gives me another reason to consider it for my own publishing.
c) Ruben Hassid: A Library Offered Alongside an Explanation
In his September 2 article about teaching AI repeatable skills, Ruben Hassid describes a library of 3,457 skills, sorted by job, which he offers through his paid subscription.
The article explains how to make skills. The library offers skills someone can download and use. That gives a reader a concrete reason to consider paying: access to prepared material organized around the work he needs to do.
You can start by collecting your recommendations in one place, organizing them by what they help someone do, and explaining why you chose each one.
4. Turn an Article Into a Practical Resource
An article that explains a method can also provide the material for a checklist someone uses while working.
Take my Citation Authority Building framework, in issue #82. It describes how an expert can interpret an emerging idea for a specific reader, with his own judgment and supporting sources. More or less what I am doing here now.
From that article, I could make a checklist for evaluating a new topic: what is the emerging idea, where did it come from, what do I think about it, and who could use this interpretation? I could put examples and sources beside each question.
But simply extracting the four steps would leave something important behind. The article also explains why interpreting someone else’s idea requires giving good credit to the originator. Without that explanation, the checklist could become advice for claiming another person’s idea as your own.
That is the difficulty with making knowledge reusable.
You have to decide what must stay with each piece so that it keeps its meaning.
5. Keep the Evidence With the Recommendation
The same problem appears when a recommendation is taken out of an article. The name of the recommended tool may survive, while the reasons and limitations disappear.
Was it tested? By whom? For what kind of work? Is the recommendation still current?
Once those details are missing, a reader has little basis for trusting it. An AI can repeat the recommendation in a confident answer without restoring any of the missing evidence.
This is why my interpretation of liquid content includes three things I would keep with the material wherever I use it:
Provenance: where it came from, who first said it, and a link to the original source. My interpretation should be distinguishable from the original claim.
Process: how I reached my conclusion, what I compared, what I left out and why.
Verification: what I actually checked, when I checked it, and what remains uncertain.
For a tool recommendation, that means saying what I tried, who I would recommend it to, its important limitations and the date of the test. If I have only read about it, I need to say that too.
At this point, the resource becomes useful because of the selection, and trustable because someone can examine how that selection was made. A list of names alone gives him neither enough help to choose nor enough evidence to judge my advice.
This is where curation, paradata and liquid content come together for me. As knowledge is reused in more forms, the record of where it came from and how it was checked becomes more necessary.
I would build that record into the resource from the beginning. It cannot guarantee that another person or an AI will preserve every detail. But it gives someone a source to go back to and check.
6. How To Start Building Resources From Your Articles
I would begin with one question that my work already helps people answer. Something a reader needs to choose, understand or do.
Then I would look through my last twenty articles and collect the parts that help answer it, keeping the sources and explanations beside them.
The form would depend on the help I want to give:
A comparison, to choose between tools or services, with criteria and limitations.
A directory, to find people, organizations or sources, with a reason for each inclusion.
A map, to understand a field, its main ideas, connections and disagreements.
A decision tree, to work through questions that lead to a suitable recommendation.
A collection of cases, to see what happened in practice, under what conditions, and with what results.
I would build a small version that answers one question well. For each entry, I would check that the source, my reasoning and the verification details are present. I would also decide how often I can realistically update it.
Then I would write an article showing how to use it: the question it answers, one useful example, and a link to the complete resource.
The next time I write about that subject, I would add whatever improves the resource and correct what is no longer accurate. That way the research I do for each article also improves something a reader can keep using.
Make Your Expertise Buyable
You may already have useful work spread across your site and your articles. How much of it would a new visitor understand? Would he see what you know, how you choose, and where you could help him?
I look at your online presence as a new visitor would. I record what I see and think, send you the video, and then we meet online to go through what is working and what I would change if I were in your place.
If you would like that outside view, check Make Your Expertise Buyable.
The essence remains this:
For centuries, sharing knowledge to demonstrate your competence and authority meant packaging what you know into something that people could read, watch, or consult: a book, a map, an archive.
We may be entering an age in which sharing your know-how also means making what you know available as pieces that people, and machines, can assemble around in real-time to satisfy needs, requests and questions you cannot anticipate now.
Paradata
Why / Intent:
I wanted to explore liquid content in relation to the work I do on curation, trust and authority. My interest is in what an independent expert can build from the knowledge he already has, and how articles could help people discover and use those resources.
Process:
I started from a feeling that there may be something worth exploring at the intersection of the emerging concept of “liquid content” and my focus on trust, authority and credibility building in the age of AI.
I had heard the term “Liquid Google” from Giorgio Tave but never explored it in depth, and just a few days ago my reader and long-time supporter Stefan Diepolder - an expert in helping companies preserve their valuable know-how - mentioned the term “liquid content” as something that really made a difference in some of his recent projects.
So I said to myself: “Wait, maybe I should stop and check what ‘liquid content’ really means and see if it a) matches something I have already explored but possibly under a different label, or b) is something interesting I don’t know anything about and which I should investigate.”
So I did some research to understand existing uses of the terms and Giorgio’s precise interpretation of the Liquid Google concept. Then I started recalling that I had recently mentioned in the newsletter a concept by Sangeet Paul Choudary that expressed the same basic idea.
*Here is the full transcript of my initial research dialogue with AI and what came out of it.
At that point I had enough data to understand that the topic was super interesting, that it was still in its emergent phase, and that it could be further interpreted and refined based on context and goals. Thus I set out to attempt to provide my personal interpretation of Liquid Content while intentionally trying to expand it in a way that went beyond what had already been written about it.
I explained the whole thing to ChatGPT 5.6 Terra, brainstormed possible interpretations and provided it with references.
I started drafting outlines of what shape this could take. I did 11 different versions before reaching something I considered a possible baseline to start from.
At that point I realized I may have accidentally found the conceptual bridge between Thinking in Public → AI-readable publishing → structured knowledge → liquid content.
I then went to Claude Opus 5 and shared the work done up to that point (all of the different steps and revisions) and asked it to work on it and bring a new draft outline that highlighted the key essential aspects of this understanding and the transformations it would bring with its emergence.
That’s when the full idea emerged: Before AI, humans or software systems had to decide in advance how those modules would be recombined.
Now an AI can do the assembly at the moment of consumption.
That’s the big shift.
Imagine somebody arrives at your body of work and asks:
What does Robin Good think an independent expert should do to become more trustworthy in the age of AI? Give me the five most important principles, show me the evidence behind each, and tell me where his thinking changed during the last year.
There doesn’t need to be an article answering that question.
An AI could construct the answer from your claims + evidence + judgments + examples + relationships + paradata.
Having now a strong draft outline combining these ideas, I went to ChatGPT 6 - Astra and worked through many more drafts until I was satisfied with the structure and presentation.
Astra produced a new draft. I liked the opening and the sections about independent experts and real examples, but the synthesis left out something essential to my interpretation: provenance, process and verification. I asked Astra to bring those into the opening, alongside the shift from articles to resources.
Finally, the section about trust still felt attached at the end. I asked Astra to show the problem that makes it necessary: when knowledge is taken out of an article, its sources, conditions and evidence can get lost. Keeping those with the material then becomes part of making the resource useful and trustable.
I re-titled the sections - I had already provided a guiding title for the whole piece - and then started to do the final editing and checks.
Times:
Brainstorming: 2.5 hrs
Outline Drafting: 2.5 hrs
Drafting: 1.5 hrs
Editing, writing, revising: 2 hrs
Formatting and links: 1.5 hrs
Paradata: 1 hr
Cover design: 1 hr
Verification: 0.5 hrs
Total apx. time: 12.5 hrs
Tools:
ChatGPT 5.6 Terra / 6 Astra - for brainstorming and drafting
Claude Opus 4.8 / 5 - for brainstorming and drafting
ChatGPT Image 2.5 - cover design execution
My Curator app
Cover image: Liquid content. Concept and design Robin Good. Executed by ChatGPT Image 2.5.
Follow a path with a heart.
From Koh Samui (TH)
Robin Good