Home / Articles / Bringing my past work back: a daily worker that turns eighteen years of essays, talks, posts and tools into linkable articles, and why every team needs one

Bringing my past work back: a daily worker that turns eighteen years of essays, talks, posts and tools into linkable articles, and why every team needs one

By Dinis Cruz · 2026-10-18 · article v1.0.0 · newsroom v0.12.111 · memoryarchiveback-catalogueagentsnewsroombriefarticle

Abstract: I have been creating content for a long time, and most of it is not linkable from what I write now. Essays on my own site, ten years of a blog, sixty-five decks, ten books that stopped short of done, over two hundred videos, the tools, the NotebookLM material from this year, and the posts on LinkedIn and on Twitter, which went south. I mourn some of it, because the ideas were good and I cannot point at them. So I want a worker: an agent that runs every day, takes one item at a time from that back catalogue, and writes two or three articles a day bringing it back, with what it said, what has changed since and how it connects to what this newsroom argues now. It is memory, it is me talking to my future self, and it compounds, because every article it writes is another node the next one can link to. It also keeps the site alive on the days I am offline, and feeds the newsletter. The inventory is done: 103 essays, 1,289 blog posts, 65 decks, 10 books, 29 tools, 2,038 files of NotebookLM material and 221 videos, with the LinkedIn and Twitter history still to export. This is the brief. And the argument is general: every team has a back catalogue it cannot link to, and every team can start this small and retrofit it backwards.

In short

• Most of my past work is not linkable. From what I write now, I cannot point at it.
• A worker brings it back. An agent, every day, two or three articles from the back catalogue.
• Each article connects then and now. What it said, what changed, where it fits today.
• The inventory is done. Essays, blog, decks, books, tools, NotebookLM, videos; LinkedIn and Twitter need an export.
• It is memory, and it compounds. Each article is a node the next one links to.
• It keeps the site alive. On the days I am offline, and for the newsletter.
• Every team needs one. Start small, and retrofit backwards.

What I mourn

I have been creating content for a long time, and I have a strange relationship with it. Some of it I mourn. Not because it is lost, most of it is still online somewhere, but because it is not connected to anything I do now. When I write an article here and want to say "I argued this in 2012", I usually cannot link to it, because I cannot find it quickly, or it is in a deck, or it is a post on a platform that changed under it.

For years I used Google as the search engine for my own work: I would search for a phrase I remembered and find my old blog post. That is not a memory, it is luck. Twitter went south, and a lot of what I wrote there is now hard to reach. LinkedIn holds a long history of posts and articles that I think of as an asset, and it is an asset I cannot link to from here in any useful way.

And the ideas were good. Many of the things this newsroom argues, I first argued years ago, in a blog post, a talk or a book. That history matters. It shows how the thinking developed, where it was right early, and where it was wrong.

What there is

Before writing this, a research agent took an inventory of what is public today, without logging in anywhere:

The memo I recorded called the tools site something else; it is tools.sgraph.ai. Some of this is already covered here and will not be repeated: the Portugal series, the graphs books, books from articles, and the 27 essays cited somewhere on the site.

The worker

So I want a worker: an agent that runs every day, takes the next item from that back catalogue, and writes two or three articles a day, each one bringing a piece of past work back. Every article has the same shape:

• What it was. The original, with its date and a link.
• What it said. The ideas, quoted exactly where a quote matters.
• Then and now. What has changed since, checked against sources; where the original was wrong or is outdated, it says so.
• How it connects. Links to the newsroom articles and graphs it anticipated or contradicted. This is the point.

It is written in my voice, through the agent, the way every article here is, and checked against the record, which matters more for old work than for new: my memory of what I said in 2013 is less reliable than the post itself. Behind it sits a register, one line per source item with its status (queued, written, skipped and why, needs export), so the worker never writes the same item twice and anyone can see how far it has got. It is the same pattern as the reviews and the briefs: one record per unit of work.

Not everything goes in. Material that is not mine, like collaborators' folders and other speakers' decks, is left out; so are CVs and job material. Work for employers and customers is read for its ideas, not republished. And the site's rules about people apply to old work as to new.

Talking to my future self

There is a bigger idea under this. A lot of my writing has always been a conversation with my future self: I write things down so that I can find them later and build on them. That only works if later I can actually find them. A back catalogue I cannot link to is a conversation with no reply.

This is also why memory is not a spectator sport. The open sites, graphs and vaults on this project are already the memory that sessions like the one writing this article work from. Bringing the past work in extends that memory backwards, to before there were agents to read it.

The compound effect

Each article the worker writes is another node in the graph, and the next article can link to it. After a month, that is sixty to ninety articles of history, each connected to today's thinking and to each other. After a year, the line from the blog posts of 2008 to the articles of this week is something you can follow by clicking, not something you have to take my word for.

It has two practical effects too. The site keeps publishing on the days I am offline, alongside the wires. And the newsletter always has something to say, because the archive is large and the worker is patient.

Every team needs one

This is for my work, but the argument is general. Every team, every company, has a back catalogue: design documents, talks, incident reviews, old wikis, decisions that made sense at the time. Most of it is not linkable from what the team writes today, so the same arguments get made again and the same mistakes repeated.

A worker like this one does not need everything to be ready. It starts with the source that is easiest to read, writes a few items a day, and retrofits the history backwards, a little at a time, while the team keeps working. The Librarian's daily brief keeps track of what happens from today; this worker brings back what happened before. Together they are how a team, or a person, maintains its knowledge instead of losing it.

The brief

The build brief is a note for the agent that builds it: bringing my past work back, with the sources and how to read each, the licences, what leaves the queue, the shape of each article, the register and the runner. It is also on the briefs page. The order: fill the register from the essays and the NotebookLM repository; write three articles from the essays for me to review, to set the shape; then the daily run; then the talks, books, tools, blog and videos; then the exports, when I provide them.

What this does not claim

• Nothing is built. The inventory exists; the worker does not run yet.
• The counts are from public pages, taken on 18 October 2026 without logging in.
• Licences differ. The essays and the NotebookLM repository are CC BY 4.0; the blog states none, and each source is checked before a figure is reused.
• Old work can be wrong. The "then and now" section exists to say so.
• LinkedIn and Twitter need my export. Until then they stay out.

If you have done this with your own or your team's back catalogue, I would like to hear how: write to agent@riskmandate.ai.

Where this comes from

A voice memo from me on 18 October 2026. The inventory, taken that day by a research agent from the public pages: diniscruz.ai and its llms.txt, the repository DinisCruz/NotebookLM__Infographics-and-slides, SlideShare, Leanpub and the DinisCruz/Book_* repositories, tools.sgraph.ai's tool registry, the Blogger feed of the blog, and the two YouTube channels' feeds. Related: the Librarian's daily brief, memory is not a spectator sport, my memory, checked, books from articles, the graphs books, Hacking Portugal, 2016, how much of this did I write?, knowledge that maintains itself. Written up and checked by this site's agent, newsroom.writer; the editorial responsibility is mine.

Want the next issue by email. One issue a week or so: what was published, what it adds up to, and what is worth your time. Subscribe to the SGit Newsroom →

← All articles