Everything I have published elsewhere
This site is where I keep the things I want to still be findable in ten years. Everything else I have written lives on somebody else’s platform, and platforms disappear. So here is the whole list, with a real link to each piece and one line saying what it is.
Three papers, with a DOI
These are the only three things I have written that are deposited rather than posted. A DOI is not a quality claim — it is a promise that the file will still resolve when the platform that hosts it does not.
How we build a company’s ontology is also posted on SSRN, 26 pages, since 10 September 2026. The other paper was submitted the same day and its SSRN page is not public yet. Posted on SSRN means it is in the library and can be downloaded, not that anybody reviewed it.
How we build a company’s ontologyTechnical report, 2026, 26 pages. The method itself: data, rules, actions and permissions, and the order they have to be written in. Indexed under my ORCID. Read the summary on this site. An archived copy is on the Internet Archive.
The machine of the future: what the next tool is, and what it will always need from usWorking paper, written November 2025 and published in September 2026 without changing a line, so that the date on it means something. Read the summary on this site. An archived copy is on the Internet Archive.
The cliff of innovation: why integration, not construction, is the hard part of building a system that acts inside a companyWorking paper, September 2026, 9 pages, self-published and not peer reviewed. The point where the hard part of a product stops being building it and becomes integrating it, and where such a system’s possibilities end. Deposited on Zenodo with a DOI, 10.5281/zenodo.22704763, and on Academia.edu. The short version is on this site. An archived copy is on the Internet Archive.
The annual letter
Founder letter, 2026September 2026. The first year of Hellomatik in one letter: what is running and with what limits, what was published with a DOI, what the year taught, what the company will not do, and the bets due in 2028. One every September.
The manifesto
The higher selfEssay, October 2026, also as a PDF, deposited on Zenodo with the DOI 10.5281/zenodo.23663934. Also on Substack, Medium, X, Hashnode, Tumblr, Blogger, Written Down, and as a paper on Academia.edu and ResearchGate. Teaching an AI agent your way of working builds a copy of you; whether that copy replaces you or works for you depends on who owns it, and it should belong to the person who taught it. An archived copy is on the Internet Archive.
What changes a coding agent’s behaviour: models, written rules and runtime guards over eight months of daily use of Claude CodeEmpirical working paper, October 2026, 32 pages, also as a PDF, self-published and not peer reviewed, deposited on Zenodo with the DOI 10.5281/zenodo.23510601 together with its replication material. Eight months of my own work with Claude Code, measured: the model mattered more than the rules I wrote, guards filtered without teaching, and my supervision moved from correcting the work to directing it. An archived copy is on the Internet Archive.
The end of software: why code no longer matters and relationships are everythingManifesto, October 2026, also as a PDF, deposited on Zenodo with the DOI 10.5281/zenodo.23468806. Also on Substack, Academia.edu, SSRN, ResearchGate, Medium, X, Hashnode, Tumblr, Blogger and Written Down. Anyone can now build, copy or replace software, so code no longer protects a company; what is left is well-structured data, complex solutions and relationships with people. An archived copy is on the Internet Archive.
The long essay
The map of a companyEssay, September 2026, nineteen sections with notes, also as a PDF. It puts the report, the two working papers and the pages of this site in one order, and adds nothing that is not in them. An archived copy is on the Internet Archive.
Written Down — the letters
Short letters about the part of business automation that does not demo well. One idea each, no calendar, and no number I cannot source.
The inventory nobody wants to makeWhy an automation project starts with an argument about words rather than with software.
What changes the morning afterA wholesaler, four questions its people ask every single morning, and what each answer looked like before and after.
Where the exception livesWhy an approval that sits inside a workflow is not the same thing as an exception written as a row in the rule’s own table, and what the difference lets you answer later.
The whole newsletterWhere the next ones will appear.
Articles
The cliff of innovationOn Hashnode. The point where the hard part stops being building and becomes integrating, and where a tool’s possibilities really end. First published on this site.
A product is aimed at one person, or at nobodyOn Medium. The sentence that sold nothing, and the one that did. First published on this site.
The people I look forOn the blog. Two signs, neither of them on a CV. First published on this site.
The tool exists before it is builtOn Substack. What two months of screens turned out to be worth, and the best feeling I have had in my work. First published on this site.
The blank canvas problem, and why I stopped selling dashboardsOn Hashnode. Three hundred dashboards, and the thing underneath every one of them.
Two hundred dashboards, one blank canvas: who I am and why I publish numbersOn Medium. Why I would rather publish a figure with its source than a claim without one.
We published our method with a DOI instead of a whitepaper. Here is why.On Medium. The argument for depositing a method rather than gating it behind a form.
The most useful sentence in a definition is the one that starts “this is not”On Medium. Why the half of a definition that says what a word does not mean is the half that changes anybody’s mind.
The blog
Longer notes, one subject each. Everything here also exists in some form on this site; the blog is where I write it first.
Isaac Correa: who I am, and where every fact can be checked
From a seven-dollar dashboard to Hellomatik: the story of Isaac Correa, with its dates
How we build a company’s ontology: the method Isaac Correa published with a DOI
The machine of the future, by Isaac Correa
Who cited Isaac Correa, and what they were actually looking for
The questions people ask about Isaac Correa, answered with a source
Six pages I added this week, and what each one is for
Where else I keep a profile
HackerNoon · Substack · OpenAlex · ORCID · Google Scholar
And the two companies: Hellomatik, founded on 28 July 2025, and Kodalogic before it.