Cover content drafts
Two drafts below: the blog manifesto post and the Substack welcome issue. Both are written in the register of your current About page. Placeholders in [brackets]. Where I left the technical lineage generic, add specific names yourself if you want them public.
A. Blog manifesto post — “Welcome to The Workbench”
Suggested slug: posts/welcome/ · pin to top of the Posts page · publish on launch day
Every workshop starts as a bench, a few tools, and a question that won’t leave you alone.
Mine is this: what happens to financial modeling when the model can explain itself — and when an AI can sit at the bench with you?
I am a professor of corporate finance. For most of my career I have taught and done research with spreadsheets — and fought their two oldest defects: formulas that say =B7*C7 instead of Revenue × Margin, and logic so scattered across cells that nobody, including the author, can audit it a month later. The multidimensional spreadsheet tradition, which began with Lotus Improv, solved much of that: formulas written over named things, models whose structure you can read. I have worked in that tradition for years, and this project carries a tribute to it in its name.
Impromptu is the tool at the center of this workbench. It is a financial modeling environment I am building that keeps what multidimensional spreadsheets got right — a self-contained model where structure, data and calculation are all visible and all named — and extends it toward what modern computational research needs: labeled-array data structures of the kind scientific computing relies on, live recalculation, notebook-style reports inside the model, and a connection to AI assistants that can read and operate the model directly.
That last point deserves a plain statement, because it is the thread that ties this site together. Impromptu was built with AI assistance, and it is built for AI-assisted work: an assistant can open a model, set assumptions, write formulas, check results and draft the report — while you watch every step on screen and remain the one who decides. Not magic, not a black box: a workbench where the machinery stays legible even when an AI has its hands on it.
What will appear here
Episodes, roughly every two weeks, each one a small self-contained piece of work: a model you can see, a video demonstration, and a written walk-through. Two series to begin with:
AI-assisted financial analysis. Real sessions where an AI assistant and I do the everyday work of finance — building a forecast income statement, analyzing financial statements, pricing a bond — inside Impromptu. What works, what fails, where judgement still earns its keep.
The ABC of language models. The series already underway here: tiny Transformer models, small enough to inspect completely, that show what attention, embeddings and next-word prediction actually do. If you use language models in management or finance, this is the intuition the demos and the hype both skip. Three long essays are already published; videos of the tiny models in action come next.
A third series, My finance class, starts with the academic year: the corporate-finance classics, each built from a blank model, for students and for anyone who wants the fundamentals with every assumption visible.
Other projects — portfolio optimization, simulation of blockchain contracts, small machine-learning experiments — are on the shelf behind the bench. They will appear when they are ready, not before.
What this is not
I am not going to tell you whether AGI is near, or what AI means for the future of work. Plenty of people are louder and more confident on those subjects. What I have is first-hand experience: an end user of frontier AI tools in daily research and teaching, and a hobbyist who builds tiny models to understand the big ones from first principles. The reports from that bench are what I can offer, and I think they are worth more than one more opinion.
Can I try Impromptu?
Not yet, by download. Impromptu is a working prototype; for now you will see it in action in every episode, and that is deliberate — I would rather show real work than ship an installation problem. A small group of early users will get hands-on access later this year; if that interests you, write to me and tell me what you would build with it.
Follow along
New episodes are announced on the newsletter — free, no paid tier, the full content always available here. Demonstrations live on the YouTube channel. Everything permanent lives on this site.
Welcome to The Workbench. The light is on.
B. Substack welcome issue — “The Workbench is open”
Send on launch day. Substack subject line: The Workbench is open · preview text: “A finance professor’s workshop: AI-assisted modeling, transparent tools, tiny language models.”
Welcome — and thank you for subscribing before there was much to subscribe to.
The Workbench is my workshop as a corporate finance professor, opened to visitors. Two lines of work run on the same bench: building transparent tools for financial modeling, and running hands-on experiments to understand how AI language models behave. They meet in one question: what happens to financial modeling when an AI can sit at the bench with you?
Today three things go live:
1. The blog — the workshop itself. The Workbench is where everything permanent lives: long essays, field notes, and from today the manifesto of the whole project. Three essays on how Transformer language models actually work — attention, embeddings, the final next-word step — are already up.
2. The YouTube channel — the demonstrations. The Workbench on YouTube opens with a two-minute tour and the first full episode: a working session where an AI assistant and I analyze a company’s financial statements inside Impromptu, the modeling tool I am building — assumptions, formulas, error correction, final report, every step on screen.
3. This newsletter — the announcements. Roughly every two weeks, when a new episode is ready, you will get a short letter like this one: what the episode shows, why it matters, and the links. That’s all. Free, no paid tier, ever — the full content is always on the blog and the channel. Your subscription is the project’s only growth engine, so if an episode is worth your time, forwarding it is the best thanks.
What’s next: the following episode opens the ABC of language models video series — a Transformer so small you can watch every weight, trained live, taken apart on screen.
If you have a finance modeling problem you would like to see treated as an episode — or you’d like to be among Impromptu’s first hands-on users later this year — just reply to this email. I read everything.
— Luca
C. Notes on both drafts
- Channel URL placeholders must be filled after the channel exists; create the channel before scheduling either piece.
- About page: after publishing the manifesto, trim the About page so the two don’t compete — About stays as the short bio + “how this is made” (the AI-collaboration statement there is excellent; keep it verbatim), and link to the manifesto for the project’s goals.
- The manifesto replaces nothing on Substack: Substack’s own About page should be two paragraphs distilled from draft B, plus the three links.
- Tone check: both drafts avoid promising specific future series beyond the three named, per the plan’s “defer the pipeline” rule.