CASK ON CASK
We run the method on our own work first.
Before we ask anyone to trust the approach with their expertise, we use it ourselves. These are five of our own projects. Each one starts with knowledge too large, too scattered or too private to use as it is, and each one has to earn its audience every week.
Our own work. None of these are client engagements.
The method, in one line
Every project below follows the same path. We take a source that is hard to use, Cask captures it, checks it and organizes it, and the output is something people can read, search or act on. What people do with the output then feeds back into the source.
SourceCaskOutput
Our own work · Markets research
01 · Staring Down the Barrel
It started as one investor's morning routine.
Sebastien has worked in finance for twenty years, and for all of them he has kept a daily research practice. Reading markets is part of the job and part of the pleasure. As his commitments grew, the reading didn't shrink. Sell-side research, company filings and the newsletters worth following add up to something close to a full-time job, and he already has one, as a venture partner.
So he codified how he reads, and handed the reading to AI. The aim was a morning note like the ones he used to write for institutional clients, with the whole pile already scrubbed and the reports that deserve a closer look brought to the top. Every point links back to its source, so when something matters he can go straight to the original and read it in full. What used to take four hours now takes about one. The work changed from sifting to taking in.
The system could run with far less of him than it does. He chooses to stay hands on, reading the sources behind the note, arguing with it and approving what goes out, because staying close to the material is how he keeps learning. The AI does the sifting. The judgment stays his.
Then came the obvious question. If the note sharpens one investor's view, who else would want it? Years of writing to other investors suggested quite a few people would. So the same process became a publication. Staring Down the Barrel started as a Substack built on distribution lists Sebastien had kept for years, and now has over 6,000 subscribers. It has since grown to daily commentary on X, a Telegram channel, Bluesky and Reddit.
Source
Sell-side research, primary disclosures such as company filings, and newsletters.
Cask
AI reads across the whole pile, scrubs and ranks it, and sources and dates every point. Sebastien reads, edits and approves before anything is published.
Output
A private morning note for one investor, and a public publication across five channels.
What it taught us. Once the way an expert reads was written down, it could serve more than one audience. The private morning note and the public publication run on the same process, and both stay sharp because the expert is still in the loop. That's the pattern we bring to clients: capture how your best people think, and it can work for your team and for the market at the same time.
Our own work · Video
02 · Short-form video
Knowledge that was only ever going to be read once.
Sebastien reads far more about business and finance than his work ever uses. How fortunes were made and lost, why famous deals went wrong, the money decisions most people get wrong without noticing. For years that reading stayed in his notes and his head, useful to exactly one person.
At the start of 2026 we began turning it into short videos for TikTok, Instagram and YouTube, made with AI from his research. They cover personal finance, business history and the less obvious corners of how money works.
Starting from zero in January, the videos have passed 2.5 million views and 10,000 followers across the three platforms.
- Views
- 2.5M+ views
- Followers
- 10,000+ followers
From zero, January to November 2026. TikTok, Instagram and YouTube combined.
Source
One person's reading in business history and personal finance.
Cask
Researched, shaped into a story, checked for facts, and produced with AI.
Output
Short videos on three platforms, and an audience that didn't exist a year ago.
What it taught us. Sometimes putting knowledge to work means changing its form completely. The same reading that sat in a notebook now reaches people who would never open a research note, because it arrives as a sixty-second story. The format changed entirely. The facts couldn't change at all.
Our own work · Public company research
03 · Research on company filings
The source. Quarterly and annual filings with the SEC. Every public company publishes them, and almost nobody reads past the headline numbers. The useful detail sits in the notes.
What Cask does. Every weekday a research desk reads the new filings for one sector, and on Sunday a summary pulls the week together, the way a CIO would read across their analysts. From that reading, the pieces worth publishing go out on @caskadvisory, each one built on what changed between filings. Before a piece is published, it states a test: what the next filing would have to show for the reading to be wrong. Then it publishes the result either way.
An example. In October we noticed PayPal's June-quarter filing used a word its previous five filings had not. On 22 October we wrote down how the September-quarter filing would test that reading. When it arrived, it named the item again but gave no figure we could score against. We published that on 28 October, said the test couldn't be settled yet, and moved it to the annual report.
What it taught us. Writing the test down before the facts arrive is uncomfortable, and it is the most useful habit we have. It keeps a reading honest, and when the answer comes back "unclear", saying so in public is what makes the next reading worth trusting.
Our own work · Expertise capture
04 · A voice, captured
The closest thing on this page to what we do for clients.
Before any of this existed, Sebastien wrote a morning note and a weekend letter for institutional clients. About 9,000 words of that writing survive from 2020 and 2021. For years they were old files, read by nobody.
We treated them as a source. We measured how he actually writes: the length and rhythm of his sentences, how he anchors a number, how he names a source, the words he never uses, the way he opens on a particular and closes forward. The result is a written guide to his voice, with the measurements and the reasons behind them.
Everything published in his name is now drafted against that guide and checked before he sees it. He still edits every piece. The difference is that he starts from something that already sounds like him.
Source
Old client notes and letters, sitting unread in a folder.
Cask
Measured, written down as rules with reasons, and tested against new drafts.
Output
A style guide that every new piece is written and checked against.
What it taught us. The most valuable knowledge was in documents nobody thought of as knowledge. Nobody would have called a stack of old morning notes an asset. Read carefully, they held a precise account of how one expert thinks on the page, and that turned out to be the hardest thing to rebuild any other way.
Prototype
05 · Cask Studio
The source. Whatever you tell it. Studio asks three questions about something you're building, including the one thing you're least sure about.
What Cask does. Turns your answers into research questions, searches public sources, and looks first for what complicates the obvious advice. Every citation is checked against what was actually retrieved, and any point it can't support is dropped before you see it.
The output. A short brief, built around your specialty, with its sources listed and its assumptions marked for testing.
What it taught us. The research is the easy part. The hard part is the first question, getting someone to say what they actually know and what worries them. Studio does that in a minute. An engagement does it over weeks, with interviews, and goes much further.
How one person runs all of this
Sebastien works full time as a venture partner. SDTB, the videos, the filings research and Cask run alongside that because his methods are written down. Each project has a runbook: what gets checked, what needs his approval, what must never repeat, and how to confirm something is done before calling it done. The work that needs his judgment comes to him. Most of the rest doesn't need him at all.
That's the same thing we help clients build. It starts with writing down how the work is actually done.
What this shows, and what it doesn't
These projects show the method working on material we chose and know well, at our own pace and our own risk. They don't show results for a client, and we don't claim any. What they do show is how we work: we name our sources, we check our output against them, we publish our mistakes and open questions, and we keep the work running long enough for it to compound.
Our editorial standards set out how we source, check and correct what we publish.
Your knowledge is a better source than ours.
Everything above was built on our own material, most of it public. The expertise inside your organization, or in your own career, is more specific, harder to find anywhere else and worth more. A first conversation costs nothing.