How I work with Claude — Francisco Lorca

01

Claude is my chief of staff, not my chatbot

1

Knows my world

Briefed once, not every time

Claude has a permanent file with who I am, who I work with, what I'm working on, and how I like things done. It reads it before every conversation.

Context

2

Writes like me

Tone, register, language

Claude has a separate style guide for emails, memos, and presentations — in both English and Spanish. It picks the right one for the task.

Voice

3

Works inside a system

Disciplined, repeatable

Every project has the same shape. Every decision is logged. Rules earn their place by repetition. The system grew by use — not by design up front.

Discipline

A chatbot answers questions. A chief of staff knows the context — and works inside a system.

02

Everything Claude knows about me lives in one place

What lives in this folder

Who I am, what I'm working on, how I like to work.
One sub-folder per project, with the project's journal, decisions, and state.
Style guides for how I write in each format and language.
Templates for memos, briefs, and decision logs.
A research library that grows over time.

Why one folder works

I never explain context twice — Claude reads it before every session.
When I update something once, every future session sees the new version.
If I open Obsidian or any markdown editor, I see the same files Claude does.

One Obsidian vault, synced via iCloud. Notes, briefs, decisions, voice, rules — all markdown.

03

Three ways to talk to Claude — same memory across all of them

1

Terminal

Where I work now

Claude Code, in the terminal. It reads and edits files on my Mac directly and runs the whole system end to end. This is where most of my work happens now.

Most use

2

Browser chat

Writing and quick tasks

Emails, memos, decks, quick questions. Still the fastest surface for drafting in my voice when I don't need file access.

Day-to-day

3

Desktop app

Always on, alongside

Same as the browser, in its own window. I keep it open next to whatever else I'm working in.

Alongside

All three read the same folder. The terminal does the heavy lifting; the browser is for quick drafting.

04

Claude writes and designs in my style — here's how that was built

1

Voice — from my emails

Six style guides

Claude scanned a sample of my real emails, memos, and decks. From the patterns it extracted six rulebooks: email, memo, presentation — each in English and Spanish.

My words

2

Word template — co-built

From samples and structure

I gave Claude McKinsey and Goldman document samples. Together we worked through the Pyramid Principle — governing thought, MECE arguments — and landed on a single Word template I now use for every memo.

My structure

3

Slide templates — co-built

McKinsey, Goldman, Hybrid

Same approach for slides. From the consulting samples I provided, three layouts emerged. I pick the one that fits the audience — board, investor, or peer.

My decks

Voice from my real emails. Templates co-built with Claude from McKinsey + Goldman samples and the Pyramid Principle.

05

Every project has the same shape — Claude knows where to look

Artefact	What it is	Discipline
State file	A short snapshot of where the project is — current status, open items, next priorities.	Slim. Always current.
Project journal	An append-only log of decisions and events, dated. The bitácora.	Never edited. Only added to.
Decisions folder	One file per substantive decision. Context, alternatives, consequences, references.	Frozen once approved.
Outputs	Drafts in current/. Finals in sent/. Replaced versions in superseded/.	State machine on documents.

State + journal + decisions + outputs. Each project, every time.

06

The rules and the trust grow by use, not by design

Patterns earn their place

A new rule starts as a "pattern observed" — a candidate.
It sits there until it shows up three times in real work.
Only then does it get promoted to canon.
Stops the rules file from filling up with rules I'd ignore.
Lives in pattern-candidates.md, with a count next to each.

A short list of trusted actions

Fix a typo. Move a displaced file to _scrap/.
Update a status. Add an entry to a project journal.
Convert a format (e.g. md → docx).
Anything outside this list — Claude asks first. Always.

I don't write rules in advance. They earn their place. Same for what Claude can do without asking.

07

A project keeps its memory between sessions — here's what that looks like

1

Open the project

Day 1, 9:00 AM

I tell Claude: "Compass today." Claude reads the state file (where we are) and the journal (how we got here). Briefed in seconds. No rehash.

Start

2

Do the work

During the session

We draft scenarios, edit the deck, agree the recommendation. Substantive decisions get logged as ADRs — frozen once approved. Sensitive numbers paused for sign-off.

Work

3

Wrap and update

End of session

One word — "wrap." Journal entries written. State file updated. Drafts moved into outputs/current/. Day 7, when I open Compass again, Claude already knows.

Close

Sample case: Project Compass — Q2 board strategy review for a portfolio company.

08

When a session closes, three things happen — at three different speeds

1

Silent

Done without asking

Routine housekeeping: drafts filed in current/, journal entries written, project state updated, memory mirrored to the vault. Things I never want to be asked about.

Automatic

2

Propose

Confirm in chat

Items Claude surfaces in the session log under "Proposed" — promote-to-sent, new ADRs, pattern-candidate updates. I confirm in chat (by number, name, or "all"); Claude applies only those.

By approval

3

Red lines

Hard stop, never silent

Investor, legal, compliance, financial figures, legal terms in deliverables, anything that gets sent or shared externally. Flagged in the session log. I confirm explicitly or it doesn't happen.

Never auto

Wrap-up isn't one button. It's a routing rule. Each item runs on its own track.

09

Two routines I built to run my day — now mothballed

Mothballed These ran on a scheduler I retired in mid-2026 — nothing fires on a timer now. The pieces still exist: I run the post-call digest by hand (process-transcripts) when I need it. Kept here because the design is still how I'd rebuild it.

Morning briefing

Pulled today's calendar from my Mac.
Listed each meeting with prep notes.
Flagged people I should LinkedIn-check.
Wrote the top of my daily note.
Ran first thing — set the day's frame.

Post-call digest

Pulls the call transcript from my meeting notes app.
Drafts a summary; I review and add before anything is written.
Files insights into the right project's journal or person note.
Appends to today's daily note for review later.

Both wrote into a single daily file. The scheduled version is retired; the digest still runs on demand.

10

What I've built — and what runs week to week

Category	Examples	How
Live websites	ormeria.com, aristeia-press.com, samantha-devin.com, this site	Cloudflare Pages, deployed from my Mac with one command
Custom skills	Doc drafter, deck builder, design lens, NDA triage, brand voice, idea generator	Cowork plugins — versioned, shareable
Scheduled automations	Nightly state update, hourly transcript archive, morning brief, deployment scan — all retired	Ran on cron on my Mac; the scheduler was decommissioned mid-2026
Code projects	Algordian (architecture), Nemo (banking RFP build), a sanctions screener, a self-hosted email tool, an ambient-music pipeline, a co-branded operator guide with an AI companion	Iterative builds — Claude in the terminal
Documents at scale	Memos, decks, briefs, board papers — bilingual	Same voice, same templates, every time
Fund AP cycle	Invoices cross-checked against signed contracts, mismatches flagged, ledger entries drafted	Contract is the source of truth — Cowork enforces the match
London property admin	Council tax, agent fees, invoices — each checked against the lease, then booked	One discipline, two contexts. Same inputs: contract + invoice + ledger
Career and board work	Board-director targeting, conversation prep, bio iterations, follow-ups	Career is just another project. Same shape, same discipline

Not just decks and memos. The same setup runs websites, skills, automations, admin, and career work — week to week.

11

A research library that gets smarter with use

How it works

I drop in an article, report, or book. Claude reads it carefully and turns it into a structured note: what the source claims, the evidence, the open questions, and where it came from.

Each pass also updates the rules for how the next ingest should work — so the library improves itself. It recently learned to distinguish between factual research and opinion pieces — each gets a different treatment.

Querying it feels like asking a research analyst who's actually read the source — not skimmed it.

Sources read in full

71

Strategy and research reports, founder interviews, investor letters, and full books — each turned into a structured note.

Company & topic pages

76

62 cross-linked company profiles, 14 topic pages, and 2 doctrine pages — the library compounds as I add to it.

I add an article or book; Claude turns it into a structured research note I can query later.

12

Claude can reach into my email, calendar, and other apps

Connected app	What Claude can do	How I use it
Email	Read inbox, draft replies in my voice	Daily review, drafting responses
Calendar	See my meetings across all calendars	Meeting prep
Meeting notes app	Pull transcripts after a call	Processing transcripts on demand
Cloud drive	Read documents shared with me	Research, due diligence
My Mac files	Read and edit files in my project folder	Notes, drafts, project work
GitHub	Read and commit code, deploy	Websites, code projects

Each connector is a permission I grant once. Claude can read or draft, but I always approve before sending.

13

Hard rules keep it safe

Rule	Why it matters
Never delete files	If something needs to go, it's moved to a holding folder. Recovery is always possible.
Never send anything	Claude drafts emails, messages, and posts. I approve and send. No autopilot.
Always show a plan first	For anything non-trivial, Claude shows me the approach before doing the work.
Match the audience's language	Spanish for Spanish recipients, English for English. Not the language of my prompt.
Stop on sensitive topics	Investor, legal, financial — Claude pauses for explicit confirmation. Always.
Secrets in Keychain, not config	API keys and tokens live in macOS Keychain. Claude never sees live credentials in chat.

Written down once. Read every session. Never softened.

14

The principle: it's a system, not a magic answer — and it grew by use

1

Built in pieces

Small parts, swappable

Each piece does one thing — a style guide, a routine, a template, a project artefact. I can change one without breaking the others.

Structure

2

Grown by use

Earned, not designed

Rules and trusted actions don't get added in advance — they earn their place after they show up three times. The system reflects what I actually do, not what I planned to do.

Emergent

3

I still hold the pen

Drafts, not autopilot

Claude does the heavy lifting. I review every output that matters, approve every send, and freeze every decision. The strongest collaborator I've worked with — and it knows its place.

Trust

The system gets better the more I work in it. The model alone wouldn't.