DJ Von Frank Company Brain
DEMO · a working rebuild of the company-brain system DJ Von Frank builds for teams — every name and number is demo data · about this demo · djvonfrank.com ↗

Custom LLM & Brain

One approved source of truth the AI reads before it writes a word — what's true, how we sound, and what's approved — wrapped in rules it cannot break, doing the production work with the team.

01 · What This Is

What this is, and how it helps you

When a company’s story changes — a rebrand, a launch, a new line — every letter, page, and phone script has to carry it accurately. The brain is the team's shared system for getting that done: it holds what's true, how we sound, and what's approved — and it does the production work alongside you. You talk to it like a colleague; it answers from the library, never from guesswork.

01 · THE LIBRARY

One source of truth

Every fact, price, claim, voice rule, and brand decision lives in one governed library. Ask anything — the answer comes back sourced. When something isn't known yet, it becomes a numbered question for the team instead of a guess in your copy.

02 · THE GUARDRAILS

Guardrails, built in

You don't have to carry the rules in your head. The system loads them every session, scans every deliverable before handover, and refuses anything the rulebook forbids. Work at full speed; the guardrails catch the slips.

03 · THE WORKHORSE

It does the heavy work with you

Ask for a letter, a page, a deck — it drafts from approved templates in our voice, saves to your Work Zone, and routes to approval. The big jobs, like the 1,200-file collateral rebrand, run as supervised agent fleets with a person approving every batch.

02 · How It Works

The operating loop

Every request runs the same seven-step line — no shortcuts for "quick" pieces. Position in the folder tree IS approval status, so nothing ships by accident.

01 ASK

The request

Anyone in the team asks for a deliverable, a fact, a plan.

02 ENHANCE

Silent intake

The prompt-enhancer scores the ask; thin requests get 2-3 sharp multiple-choice questions.

03 ROUTE

The constitution

CLAUDE.md's routing table opens exactly the right rules and facts. Never crawls, never wanders.

04 GROUND

Facts by authority

Pricing from one file. Claims from one file. Trust order: Approved > Facts > Work Zone. Everything else is UNKNOWN.

05 BUILD

Adapt, never invent

Skills copy a gold-standard reference and adapt it. AIs diverge when designing, converge when adapting.

06 SCAN

The gate

The pre-delivery scanner: banned terms, leak tripwires, dash policy, raw hex. Exit code, not opinion.

07 APPROVE

The chain

Work Zone → review queue → Jordan* → Avery*. Only a human promotes to Approved. Nothing is ever deleted.

Trust order Approved Facts Work Zone everything else = UNKNOWN → filed as a numbered Q question
03 · Inside the Brain

The component map

Every layer, what it holds, and what it's waiting on. Cards link straight into the live documents; the burger menu opens the working views.

Governance

The Constitution

CLAUDE.md: 8 sections the AI loads every session — identity, load order, the routing table, trust order, ten hard rules, the output checklist, and the approval roster. A generated AGENTS.md mirror serves any other AI.

10 hard rules · 17-row routing table · regenerated, never hand-edited
Rules · 33 files

How to act

Guardrails (12 hard limits), master thinking-order, house style, the voice trio (brand voice, approved phrases, banned phrases with the internal/external message wall), 7 writing tones, 4 target audiences, 5 channel styles, and 6 output-type direction files that script entire builds.

7 tones · 4 audiences · 6 build playbooks · message wall enforced
Facts · what's true

The fact layer

Company overview, two authority files (pricing and claims — the only sources a number or credential may come from), the launch layer (Day-1 playbook, retention engine, media & PR plan), the decisions log, team + escalation tiers, competitors, and distilled meetings.

2 authority files · 3 launch playbooks · every fact sourced
Skills · 28 folders

What it can do

15 generation skills (decks, emails, microsites, price sheets, welcome kits, one-pagers…) each built to copy a gold-standard reference — activating at brand-lock. Plus the CEO presentation style, the prompt-enhancer front door, and the quality-control script suite.

27 skills (18 gated on brand-lock) · 45 agents on the team, 19 running
Brand · Atlas v3.0

The Brand OS

Six named colours, each shipped with the contrast ratio it was measured at and the failures written down as laws — no red, no text-legal orange on light, no seventh value. One mark redrawn for every size it has to survive, five approved lockups, and four type faces with one duty each. Delivered as a standalone document the whole system re-renders from when a single value is repointed.

6 colours · 5 derived formulas · 5 lockups · 4 icon drawings · 4 type faces contrast measured 2026-08-13
Launch copy · 7 pieces

The send-ready library

Every drafted launch piece filed verbatim from the GTM: CEO all-staff emails, customer letters (new accounts + existing accounts), the 5-email nurture sequence, crisis call scripts, the press release draft, rep talk tracks, vendor letters. All approval-gated, all slot-marked for open decisions.

7 pieces · approval + endorser-line gate on every send
Work Zone → Approved

The approval lifecycle

Eight personal zones (Jordan*, Avery*, Sami*, Morgan*, Frankie*, Dana*, Gale*, Marley*), five simple folders each: My Space, Active Projects, Send for Approval, Save for Later, Archived. The review queue holds 6 launch drafts, every one fleet-verified line-by-line before human review. Approved/ holds canonical truth in 12 categories and fills only by promotion.

8 zones × 5 folders · 6 drafts in review · QA'd 2026-08-08
Registers · always current

Memory & accountability

A mandatory changelog, the session work log, the setup-question ledger (57 tracked — the brain files questions instead of guessing, and the Questions page is answerable in place), the dated roadmap, and the rebrand migration plan with real counts.

57 setup questions tracked · every change dated · nothing ships unlogged
Coming layers

Hub, plans & reference

The team Hub (11 pages + a branded tool suite, 10 people seeded in its data model) waits on launch day by design. Three strategic plans map Apex parity and the migration. A 5 GB Apex reference library feeds the rebrand — local-only, never uploaded.

Hub gated on launch · 3 plans active
04 · The Engine

Graph engineering: the wide-work engine

The newest organ, installed Aug 8. The brain doesn't just write — it runs coordinated fleets of agents on work too wide for any single session, with independent verification wired into every run.

THE JOB DRAFTS REBRANDS AUDITS RESEARCH FAN OUT · ALL AT ONCE THE RULES FACTS VOICE THE SCANNER FRESH EYES · NEVER THE WORKER'S CHAT ONE VERIFIED ANSWER COUNTED · GUARDED · HUMAN-GATED

Fan out. Verify with fresh eyes. Merge one answer.

Wide jobs split into parallel workers — one per file, per draft, per partner kit. Every result is then attacked by independent verifier agents that never saw the worker's reasoning, across three lenses: rules, facts, and voice.

Every run counts its returns (a silent failure can't hide in a complete-looking report), anchors on signals that can't be argued with — the scanner's exit code, the authority files — and ends at a human gate. Fleets do the breadth; Jordan* and Avery* keep the judgment.

every spawned agent inherits the constitution — a worker that never read the rules is a leak vector
SPEC 1 · FLAGSHIP

Migration batch sweep

The 1,200-file Apex→Custom LLM rebrand, one partner kit per run: parallel piece-by-piece rebuild, triple-lens verify, Jordan* reviews one batch report.

STAGED · FIRES AT BRAND-LOCK
SPEC 2 · PROVEN

Review-queue QA fleet

Three fresh verifiers per draft plus the scanner anchor. Ran live Aug 8 — now a saved one-line command.

LIVE · 18/18 FIRST RUN
SPEC 3

Ingestion cross-check

New strategy doc arrives → parallel checkers diff it against every fact area; conflicts become numbered questions, never silent overwrites.

READY · NEXT KNOWLEDGE DUMP
SPEC 4

Acceptance fleet

The brain's health check run cold: parallel probes on the four proof questions plus refusal tests that must refuse.

READY
SPEC 5

Pre-upload sweep

Four parallel certifiers before anything ships: twin freshness, scanner sweep, link audit, the rules grep gates.

READY
SPEC 6

Register coherence audit

Fresh eyes diff every status surface against every other — including the rules files against the constitution itself.

READY · PERIODIC
Proof, not promise · first live run · Aug 8, 2026

The fleet's first job: audit our own launch drafts. It caught what everything else missed.

18/18
verifier verdicts returned
6
high-severity catches
3
lenses per draft
1
invented SLA found & flagged

Eighteen independent verifiers swept the six launch drafts and found what serial review hadn't: a customer-facing "48-hour executive response" commitment that exists in no approved source, production notes sitting in shippable copy, and unapproved superlatives inherited from strategy docs. Same day, the engine audited the brain itself — and caught two stale rules and a broken check in the scanner, all fixed within hours. The system finds its own faults before the market can.

The two-week standup · day by day

Plan · The Calendar

Every engagement runs the same dated line: ten working days from kickoff to finalization. This is the build sequence The Process describes, laid out on one calendar.

Build Brand Verification Finalization

The Standup · Weeks 1–2

→ DAY 10 · FINALIZATION
Sun
Mon
Tue
Wed
Thu
Fri
Sat
01
Kickoff · charter, the six zones, the registers
02
The constitution: ten hard rules + the routing table
03
Fact ingestion · the two authority files (pricing, claims)
04
The voice trio · audiences + writing tones
05
The scanner wired · every handover gated
06
Work Zone lifecycle · position is status
07
Brand OS swap · tokens locked
08
Generation skills built on gold standards
09
Agent team staged · QA fleet first live run
10
FINALIZATION · acceptance test + team walkthrough + handoff

The narrative version of these ten days lives on Build & Training, and the full dated plan is the project plan document.

27 skills · 9 running · 18 gated on brand-lock

The Skill Library

A skill is one packaged ability — a recipe the brain follows every time so the output is consistent. You don't type commands for these: open a session at the brain root, ask for the thing, and the constitution routes to the right skill. The eighteen generation skills are built and waiting — the day leadership locks the brand, their gold standards get built from the approved Brand OS and they all switch on at once.

Working skills Active now. These run the brain's daily machinery and every request passes through some of them.

graph-engineeringACTIVE

The Wide-Job Engine

Runs work too big for one sitting as coordinated agent fleets — fan out, verify with fresh eyes, synthesize one answer. Powers the QA fleet, the coming migration batches, and the brain's own audits.

USE"Run the qa-fleet workflow" · "Run this ingestion as a graph"
prompt-enhancerACTIVE

The Front Door

Runs first and silently on every request; scores completeness and asks only real-option questions before any work starts.

USEAutomatic — every request passes through it.
pre-delivery-scanACTIVE

The Scanner

The last gate before any handover: banned words, the rules tripwires, pricing sources, phone formats, dash police. Half deterministic script, half judgment. Exit code, not opinion.

USEAutomatic on every deliverable · "Scan this file"
quality-controlACTIVE

The Script Suite

The generators that build the machinery itself: the team site, the styled twins, the Asset Directory, AGENTS.md, the brain audit. Operator surface.

USE"Rebuild the guide" · "Regenerate the asset directory"
twin-syncACTIVE

The Twin Keeper

Keeps every markdown file's styled html twin current: staleness scan, regeneration with the day's stamp, and the never-hand-edit-a-twin rule.

USE"Run twin-sync"
file-owners-answerACTIVE

The Answer Filer

Turns a team member's answer into filed fact with attribution, cleared placeholders, a checked-off register question, and a refreshed site. Also lives as the /file agent.

USE"File this answer: Q-24 — our locations are..."
deploy-previewACTIVE

The Preview Bundler

Builds a gate-verified preview bundle of the site when push-to-deploy is not available, so a change can be reviewed before it ships.

USE"Build a preview bundle"
folder-cleanupACTIVE

The Tree Groomer

The standard library-cleanup process, ported from Apex's proven pattern: version ranking, archive-never-delete, card refreshes, promotion mechanics. Run on any Approved folder that has grown messy.

USE"Clean up the BioSummit folder"
karam-presentation-styleACTIVE

The CEO Deck Style

Cass Louis Kai*'s presentation language — density doctrine, photo treatments, badges, stat cards, stair-steps — overlaid on any deck bound for his desk. Ported from the proven Apex build.

USE"Build this deck in the Kai* style"

Generation skills Built and GATED on brand-lock. Each one copies a gold-standard reference that gets created from the approved Brand OS the day leadership locks the brand — then all of these switch on at once.

custom-llm-deckGATED

Branded Decks

Slide decks on the locked brand system, from the Brand OS deck theme.

USE"Build a deck about..." (after brand-lock)
custom-llm-emailGATED

Emails

Intro, follow-up, delivery, and sequence emails in brand voice and skin.

USE"Draft the welcome email" (after brand-lock)
custom-llm-executive-summaryGATED

Executive Summaries

One-page executive summaries, print-true, in the brand system.

USE"Summarize this for leadership" (after brand-lock)
custom-llm-landing-pageGATED

Landing Pages

Standalone landing pages on brand tokens, ready for hosting.

USE"Build a landing page for..." (after brand-lock)
custom-llm-micrositeGATED

Partner Microsites

A partner's full microsite — the migration's per-partner web presence.

USE"Build the Trellis microsite" (after brand-lock)
custom-llm-price-sheetGATED

Price Sheets

Partner price sheets from the filed price lists, in the brand system, as HTML templates the brain can regenerate.

USE"Build the Fairline price sheet" (after brand-lock)
custom-llm-print-one-pagerGATED

Print One-Pagers

Print-ready one-pagers: company, product, or program.

USE"One-pager on digital dentures" (after brand-lock)
custom-llm-quick-start-guideGATED

Quick-Start Guides

Step-by-step quick-start guides for partners and products.

USE"Quick-start for the scanner program" (after brand-lock)
custom-llm-rx-lab-slipGATED

RX Lab Slips

Partner RX lab slips, the working forms of the lab relationship.

USE"Build the lab slip for..." (after brand-lock)
custom-llm-socialGATED

Social Sets

Social graphics plus captions across platform formats.

USE"Social set for the announcement" (after brand-lock)
custom-llm-welcome-flyerGATED

Welcome Flyers

The partner welcome flyer — first-touch print for every onboarding.

USE"Welcome flyer for..." (after brand-lock)
custom-llm-whats-needed-guideGATED

What's-Needed Guides

The checklist guide that tells a practice exactly what to send for each case type.

USE"What's-needed guide for full arch" (after brand-lock)
brand-deckGATED

Brand OS Deck Treatment

Branded deck straight from the Brand OS deck theme tokens.

USEPart of the Brand OS treatment set (after brand-lock)
brand-emailGATED

Brand OS Email Treatment

Branded email set — intro, follow-up, delivery, newsletter, sequence.

USEPart of the Brand OS treatment set (after brand-lock)
brand-leave-behindGATED

Leave-Behinds

Print leave-behind or flyer, two pages max, on the brand system.

USEPart of the Brand OS treatment set (after brand-lock)
brand-one-pagerGATED

Brand One-Pagers

Company or topic one-pager, print-true, from the Brand OS.

USEPart of the Brand OS treatment set (after brand-lock)
brand-proposalGATED

Proposals

Client-specific proposal from the Brand OS proposal treatment.

USEPart of the Brand OS treatment set (after brand-lock)
brand-socialGATED

Brand Social Treatment

Social graphics and captions in eight platform formats from the Brand OS.

USEPart of the Brand OS treatment set (after brand-lock)
482 files · 308 folders · 244 done · 137 started · 130 planned · 96 not in demo

The Asset Directory

Every folder and file in the brain, mirrored on the Apex structure and coded by the signal pair: Landfall is built and ready, Meridian has an md explanation or sits in the Work Zone, and planned but not started carries no colour at all — nothing else gets one. Search it, browse it, open anything — every document opens its styled page.

0 files0 folders0 done0 started0 planned
Step by step · the whole engagement

The Implementation Plan

Four phases from first contact to a trained team. The two-week calendar covers phase 3 day by day; this page is the whole line, in order.

DEMO. This is an example of what the implementation plan looks like in production — in a live engagement each step carries its owner and its date.

PHASE 1 · INTAKE DONE IN A WEEK OR LESS

Send the intake form

Eleven sections; complete it incrementally — blanks become tracked questions, not blockers.

The packet compiles

Every answer keyed to its destination file; an upload manifest; a completion percentage.

Sources filed verbatim

Price lists, decks, SOPs land in source-docs/ untouched — the brain quotes them, never rewrites them.

Open questions seeded

Every unanswered field becomes a numbered item with an owner on the Team Questions page.

PHASE 2 · THE BRAND OS RUNS ALONGSIDE PHASE 1

Clone the master

Twenty collateral treatments, five chapters, zero hardcoded values — the starting point every time.

Run the Swap Protocol

The client's palette, type, marks, and name replace the token table; everything re-renders.

Audit + sign-off

Contrast measured, exclusions registered, leadership signs. PROPOSED until locked.

PHASE 3 · THE BRAIN BUILD TEN WORKING DAYS · SEE THE CALENDAR

Days 1–2 · the skeleton

Charter, the six-zone tree, the registers, the constitution with its ten hard rules.

Days 3–4 · the truth

Fact ingestion, the two authority files (pricing, claims), the voice trio, audiences.

Day 5 · the gate

The scanner wired into every handover — banned terms, policy tripwires, slot checks.

Days 6–8 · the system

Work Zone lifecycle, the Brand OS tokens wired, generation skills on gold standards.

Day 9 · the machinery

The agent team staged on real gates; the QA fleet's first live run.

PHASE 4 · FINALIZE & TRAIN DAY 10 · THE HANDOFF

The acceptance test

The brain answers its proof questions cold — and refuses the things it must refuse.

Team walkthrough

Three sessions: the tour, first drafts through the approval loop, the machinery.

Handoff

The team owns the flow; the gates that pace everything after (brand-lock, launch, phase 2) take over.

The day-by-day version of phase 3 is the calendar; the narrative is Build & Training.

Copy · fill the brackets · paste

Prompt Library

Tested against this brain. Every prompt below does what it says, including the three that prove the guardrails by refusing.

What the brain can do

Quote the written retention terms exactly, with the source file
Draft a doctor letter in our voice from an approved template
Refuse anything external before launch
Turn an unknown into a numbered question, never a guess
Run 21 verifier agents over the launch drafts at once
Rebrand a whole partner kit in one supervised batch
Answer who decided what, and when, from the decisions log
Build a deck in the CEO's presentation language
Scan every deliverable for leaks before it ships
Recall any meeting, distilled, with its action items
Keep the internal story and the external story from crossing
File a teammate's answer and update every register
Produce the Day-1 hour-by-hour cascade on demand
Write in seven approved tones, never off-voice
Track 1,200 files of collateral through the migration
Audit itself and log every change, dated

The tested prompts

The playbooks and the machinery

Guides & Tools

Everything that explains the system, runs the launch, and keeps the brain honest.

The launch playbooks

How the system works

Operating Guide

The zones, the rules, the workflow: how to work with the brain day to day.

Open ↗
The Constitution

CLAUDE.md: the contract the AI reads first. Hard rules, routing, trust order, governance.

Open ↗
Project Plan & Status

Prepared for Avery*: where we are, what's done, what the team owes, milestones ahead.

Open ↗
The Brain, zone by zone

The structure the AI navigates:

01 The Brain/ — the Custom LLM LLM itself ├── CLAUDE.md · AGENTS.md — the constitution every AI session reads first (+ its generated twin) ├── README · ROADMAP · QUESTIONS-FOR-THE-TEAM — orientation, the dated line to launch day, the open-questions register ├── Rules/ — how the AI acts: guardrails, voice + tones + audiences + channels, house style, output playbooks ├── Facts/ — what's true: company, pricing + claims authority files, team, competitors, decisions, meetings, source docs ├── Skills/ — what it can do: the request gate, the quality scanner, 12 collateral-generation skills (activate at brand-lock) ├── Templates/ — patterns: emails, LinkedIn, sales, the portable AI brand kit, the 16-prompt starter library ├── Brand/custom-llm/ — the design authority: the brand sheet (six named colours plus the derived formulas), the Brand OS v3.0 Atlas, how-the-brain-uses-it, the superseded directions ├── Work Zone/ — per-person pipelines (Avery* · Jordan* · Sami* · Morgan* · Frankie*) + shared projects; position = status ├── Approved/ — canonical truth: the full partner-category library, filled only through review ├── Archive/ — nothing is ever deleted; superseded work lands here dated ├── Log/ — the changelog (every update, dated), work log, generated asset directory └── Hub/ — the operating guide (how the team uses the brain) + the team-hub plan (activates post-announcement)

The quality machinery

Pre-delivery scanner v1.1

Every deliverable passes the automated gate: banned words, brand rules, the rules tripwires, dash police. Exit code, not opinion — hardened Aug 8 after the engine audited its own scanner.

The acceptance test

Four questions prove the brain in 3 minutes, including two deliberate refusals. Last run Aug 8: PASS 4/4.

The QA fleet Saved command

21 fresh-context verifiers sweep the review queue on demand: "run the qa-fleet workflow." First live run Aug 8.

Changelog

Every change to the brain, dated and explained. The audit trail — now a tab of its own.

Open →
DJVF Brand OS · v3.0 · Atlas

The Brand OS

The design system this demo is built in. Six colours, four faces, one mark redrawn at every size it has to survive. Every ratio below was measured on 2026-08-13, and the ones that fail are written down as laws rather than nudged away.

DEMO BUILD. This page shows the system the demo itself is built in, so the components on every other page can be checked against it. In a real engagement it carries the client’s proposed system instead, with its approval state and the open questions gating it.

The six, each with its measured ratio

Three do the work: Midnight Green grounds everything, Paper White carries the type, True North is the only warm thing in the system. Two are reserved for signal and are never decoration. The sixth is what the material falls back to when it cannot render.

#00272C
Midnight Green

Primary · ground · ink. The one dark. Heroes, headers, footers, dark bands, and every line of body type on a light page. There is no second dark. White on it 15.8:1 ✓ · on white 15.8:1 ✓

#FF6B1C
True North

Secondary · the needle. Named for the point it colours. Primary buttons, the eyebrow on dark, the live rule. Used once per view, twice at most. On Midnight 5.6:1 ✓ · on white 2.8:1 ✗

#FFFFFF
Paper White

Base · card · knockout. True white, never off-white. Cards, sheets, knockout marks, and all type on dark grounds. On Midnight 15.8:1 ✓

#09E85E
Landfall

Signal — arrived, shipped, passing. Land sighted: builds that pass, work that shipped. Never a decorative green. On Midnight 9.6:1 ✓ · on white 1.7:1 ✗

#279AF1
Meridian

Signal — in progress, information. A line of longitude: the thing you are still travelling along. On Midnight 5.3:1 ✓ · on white 3.0:1, large only

#245D63
Surface Steel

Material fallback · link · label. What the material collapses to with no WebGL. Doubles as the link and eyebrow colour on light, and draws the bezel and grid inside the mark. On white 7.4:1 ✓ · on Midnight 2.1:1 ✗

Derived — formulas, not new colours

#F4F7F7
--page

5% Steel on white. The page beneath the cards, so white sheets have something to sit on.

#EDF3F3
--band

8% Steel on white. Chips, code fills, quiet washes.

#476467
--muted

72% Midnight on white · 6.4:1 ✓. Secondary text.

#597276
--faint

65% Midnight on white · 5.1:1 ✓. Labels and metadata — the quietest legal level.

#A8B6B7
--dim-d

66% white on Midnight · 7.6:1 ✓. Dimmed type on dark — shown here on its own ground, because that is the only ground it has.

Every one of these is a mix against a named token, never a pasted hex. Repoint Midnight Green and the whole page re-renders correctly — that is the test this table exists to pass.

The laws (measured, hard)

True North

Never type on white at any size. On a light ground it is a fill, a rule, or a shape — nothing else. At 5.6:1 on Midnight Green it is legal for body text on dark, and that is the only place it gained. One True North button per view; a second primary means the first one was not primary.

Surface Steel

A light-ground colour only. 7.4:1 on white makes it the link, the eyebrow and the label; 2.1:1 on Midnight Green makes it illegal as type on dark, where Paper White carries the reading instead. Links get a 1px underline, 3px clear of the baseline.

The signal pair

Landfall and Meridian only ever mean something: status, and fills in data — bars, nodes, stat tiles, never the axis and never the label. Neither appears in a lockup, a header, a button, or anything adjacent to the mark. On light the dot carries the colour; on dark the label may. In greyscale they separate at 0.59 and 0.30 against Midnight’s 0.02, which is why the pair can carry status without changing shape.

There is no red

The palette has no error colour, because True North means action, not failure. An error is carried by a 2px Midnight Green border, the glyph, and a line of Plex Mono under the field. Focus is a 2px Surface Steel ring, never a glow. Depth is a hairline and a 1–2px lift; shadows exist only for genuinely floating surfaces.

Four faces, one family tree

Saira Wordmark only

Black 900, width axis pinned to 125 — Archivo’s blockier sibling from the same foundry, so the wordmark and the headlines share a skeleton without repeating each other. Never a headline, never a heading, never live text in production. Outline it and it becomes artwork.

Archivo Display

500–900. Headlines, section heads, card titles, buttons. 800–900 for display, 700 for anything under 24px. Headlines set at -0.025em; sub-heads are Archivo 800, never bold body.

IBM Plex Sans Body / UI

400–700. Everything a person reads for longer than a glance. Around sixteen pixels over a 1.55 leading, holding a 66-character measure, and it never competes with the figures beside it.

IBM Plex Mono Mono

400–600. Eyebrows, figures, IDs, file names, code. Tabular figures always, so a column of numbers lines up and a paragraph full of percentages does not ripple.

The mark — one drawing per size, never a scaled-down primary

A compass rose set over a gridded globe: the globe is the territory, the rose is the read on it. One point of that rose, north, is True North orange, and it is the only warm thing in the mark. The handoff between drawings is a measurement, not a preference. Clear space is 0.5 × icon height on all four sides, measured from the outer bezel, and drops to 0.3 in dense chrome — never below.

Never: rotate the mark · recolour the north needle · set the wordmark in Archivo, or in Saira at any width but 125 · put the icon to the right of the wordmark · scale the primary icon below 48 px instead of switching drawings · add a stroke, glow or shadow the mark did not ship with. Below 240 px the tagline is dropped, not shrunk; below 168 px the wordmark is dropped and the icon stands alone.

45 specialists · 19 running now

The Agent Team

A skill does one thing; an agent chains skills into a finished outcome, and each one has exactly one thing it refuses to do. That last part is the product. Checkers check, makers make, librarians remember, operators run the machinery — and every one of them obeys the constitution: nothing external, nothing invented, nothing into Approved without a human.

Say this first.The departments are new. This brain partitions the same work three other ways — by behavior, by system layer, and by deliverable type — and the word “department” appears nowhere in it. The shelf is new packaging; everything hanging on it is real and cited.

To activate one: open Claude Code in the Custom LLM Brain folder, type the slash command (it autocompletes), and describe what you need. The specialist runs with fresh eyes and reports back.

The honest ledger Counted from the roster below, never typed above it. 19 of 45 have actually run.

19ACTIVEHas a dated run in this brain. It has actually executed on real work.
7BUILTWritten and complete. It has never been run.
10STAGEDFinished and switched off, waiting on a named gate: brand-lock, launch, or partner facts.
7BUILDABLEThe rule or register it reads already exists. No agent owns the job yet, and nothing blocks it.
2BLOCKEDNeeds something that does not exist yet. The dependency is named on the card.

D1 · Office of the Chief of Staff Knows where everything stands, chases who owes what, and keeps the team able to run this after the build is over. Removes the need for: the Monday status meeting, and the training day that happened once and was never written down.

/statusACTIVE

The Reporter

Reads every register and returns the honest one-page where-are-we: done, in review, blocked and on whom, plus the one decision that unblocks the most.

TRY/status

Will never: Round progress up to sound better. Blocked is blocked.

/guideACTIVE

The Enablement Coach

Generates the operating guide and the prompt gallery from the brain itself, then runs the four-question health gate at handoff. It writes the page from the confusion it actually watched happen.

TRY/guide for the retention team

Will never: Call a handoff done on a failing acceptance test.

/chaseBLOCKED

The Chaser

Owns the open-questions register: who owes what, since when, and what it holds up.

BLOCKED: The register stores owners as roles rather than people and carries no dates. That schema change comes first.

TRY/chase everything owed by Avery*

Will never: Answer on the owner’s behalf just to clear the queue.

/cadenceBUILDABLE

The Governor

Wakes standing sweeps on schedule and caps what a fleet may spend before it fans out.

TRY/cadence weekly register sweep

Will never: Raise its own cap.

D2 · Intake Turns an ask into a scoped, owned, routed job before a single file is opened. Removes the need for: the three-email round trip that starts with “can you make me a flyer”.

runs silentlyACTIVE

The Front Door

Scores every incoming ask before any work starts. A thin one gets two or three sharp multiple-choice questions rather than a confidently wrong draft.

TRYmake me a one-pager

Will never: Let a vague ask through as though it were specific.

/intakeBUILT

The Discovery Lead

Runs the eleven-section intake. Every blank becomes a numbered question with an owner instead of a gap somebody fills in later from memory.

TRY/intake for the Trellis account

Will never: Fill a blank itself.

D3 · Knowledge and Records Files what is true, cites it on request, and says UNKNOWN rather than guessing. Removes the need for: institutional memory walking out the door with a departure.

/factsACTIVE

The Librarian

Answers anything from the brain’s filed knowledge, with a file citation for every fact. It would rather hand you a visible blank than a plausible invention, and it drafts the question that would close the gap.

TRY/facts what are our written retention commitments, exactly?

Will never: Fill a gap with general industry knowledge. The brain’s files or UNKNOWN, nothing else.

/scribeACTIVE

The Scribe

Files a meeting the brain’s way: verbatim source stored, decisions and action items distilled, conflicts with the decisions log flagged rather than silently resolved.

TRY/scribe [paste today’s call transcript]

Will never: Record a decision nobody actually made.

/fileACTIVE

The Registrar

Takes an arriving answer and lands it everywhere it belongs in one pass: the fact recorded with attribution, every placeholder it unlocks filled, the question checked off, the changelog line written.

TRY/file Q-13 is answered: Avery* confirmed the type pairing today

Will never: File an answer whose source is unnamed.

/ingestBUILT

The Cross-Checker

Diffs an arriving document against every filed fact. Conflicts become numbered questions, never silent overwrites.

TRY/ingest the updated services sheet

Will never: Resolve a conflict by picking whichever document is newer.

/marketBUILDABLE

The Market Watch

Keeps the competitor and category picture current, sourced and dated.

TRY/market who else sells into this category?

Will never: Present an inference as a finding.

D4 · Brand and Design Owns the tokens, marks and treatments every other department inherits. Removes the need for: the “is this our blue” round on every single piece.

/brandcheckACTIVE

The Brand Guardian

Checks anything visual or branded against the token system, the surface law, the mark rules and the name rule.

TRY/brandcheck 05 Work Zone/review-queue/website/info-page-DRAFT.md

Will never: Redesign the piece or pick new colors. It reports.

/brandosBUILT

The Brand OS Builder

Clones the master system, runs the swap protocol, and produces the full client system across all twenty treatments.

TRY/brandos build the Custom LLM system from the approved direction

Will never: Publish a system leadership has not signed.

/brandauditBUILDABLE

The Brand OS Auditor

Audits a finished brand system against the five-part contract before a brain is assembled on top of it.

BUILDABLE: It exists today as a document somebody runs by hand rather than a repeatable command. The packaging is the whole build.

TRY/brandaudit the current system

Will never: Pass a system with unmeasured contrast.

/migrateSTAGED

The Treatment Swapper

Re-renders the collateral library into the new brand, in reviewed batches. Runs after the Migration Lead, never before it.

STAGED: Gated on brand-lock.

TRY/migrate Trellis (after brand-lock)

Will never: Run a batch nobody approved.

D5 · Production Studio Adapts a gold standard into a finished piece, into your Work Zone, never into the Approved library. Removes the need for: the per-piece production half of an agency retainer.

/produceBUILDABLE

The Studio Producer

Routes a production ask to the right skill and the right gold standard.

TRY/produce something for the Trellis welcome pack

Will never: Invent a format that has no reference.

/draftACTIVE

The Drafter

Builds a new piece the brain’s way: right template, right tone, facts from the authority files, scanned clean, saved to your Work Zone.

TRY/draft a one-page internal update on where the brand sprint stands

Will never: Invent a fact, price, name or testimonial to fill a gap.

/revoiceACTIVE

The Voice Editor

Rewrites any copy into an approved tone, showing before and after. Fixes banned words, hype and off-voice phrasing without touching the facts.

TRY/revoice 05 Work Zone/Jordan/01-scratch-pad/partner-email.md into reassuring-advisor

Will never: Alter a fact, price or commitment while rewording.

/deckSTAGED

The Deck Builder

Builds presentations on the configurable executive overlay.

STAGED: Gated on brand-lock.

TRY/deck the quarterly partner review (after brand-lock)

Will never: Use a person’s presentation style outside its declared audience.

/printSTAGED

The Print Desk

One-pagers, flyers, trifolds, price sheets, mailers and certificates.

STAGED: Gated on brand-lock.

TRY/print the Trellis one-pager (after brand-lock)

Will never: Ship a piece with an unresolved slot still in it.

/pageSTAGED

The Web Producer

Landing pages, microsites, blog treatments and share cards.

STAGED: Gated on brand-lock.

TRY/page a landing page for the launch (after brand-lock)

Will never: Publish outside the approval chain.

/socialSTAGED

The Social Desk

Frames, stat cards, quote cards and announcement sets.

STAGED: Gated on brand-lock.

TRY/social an announcement set for the rebrand (after brand-lock)

Will never: Post anything, anywhere. It produces files.

D6 · Communications Owns the moments Apex Instruments* speaks at scale: sequences, launches, and the script for the day it goes wrong. Removes the need for: the PR retainer, and the crisis plan that exists only as a conversation.

/commsBUILDABLE

The Communications Lead

Owns the launch sequence and who hears what, in what order.

TRY/comms plan the rebrand announcement

Will never: Let an external piece go out before its internal one.

/nurtureSTAGED

The Sequence Writer

Multi-touch email sequences built from the approved library.

STAGED: Gated on brand-lock.

TRY/nurture a five-touch sequence for lapsed accounts (after brand-lock)

Will never: Send. It drafts only.

/announceSTAGED

The Announcer

Launch kits: press draft, fact sheet, talk tracks and FAQ.

STAGED: Gated on launch.

TRY/announce the Custom LLM launch (after the date is set)

Will never: Publish before the date.

/crisisBUILT

The Crisis Desk

The pre-approved scripts for the day it goes wrong, one per scenario.

TRY/crisis a shipment defect reaches a partner

Will never: Improvise a statement.

D7 · Revenue and Accounts Everything per-account rather than per-company: briefs, proposals, price sheets, onboarding kits. Removes the need for: the sales-ops hire nobody made, and the week-long proposal.

/briefSTAGED

The Account Briefer

The one-page brief on any account: history, contacts, terms, collateral, open items, and who owns it.

STAGED: Gated on partner facts.

TRY/brief Trellis (after partner facts land)

Will never: Brief from memory rather than from the file, or pad with industry generalities.

/proposalSTAGED

The Proposal Writer

Scoped proposals with every figure traced to the pricing authority.

STAGED: Gated on brand-lock.

TRY/proposal BioSummit (after brand-lock)

Will never: Quote a number that is not in the authority file.

/priceACTIVE

The Price Desk

Answers any pricing question from the single legal source, with the citation attached.

TRY/price what is quotable on the standard tier?

Will never: Quote a price from a meeting note.

/kitSTAGED

The Kit Assembler

Per-partner onboarding kits cut from one master architecture.

STAGED: Gated on brand-lock.

TRY/kit BioSummit (after brand-lock)

Will never: Ship a kit carrying a partner’s mark it has not been given.

/onboardBUILT

The Onboarder

The first-ninety-days pack promised at signing, built in one consistent run.

TRY/onboard Trellis

Will never: Promise a date operations has not agreed to, or leave another partner’s details in via template residue.

D8 · Quality, Risk and Assurance The only department that can say no. Owns the gates, the authority files, the registers, the fleets and the incident loop. Removes the need for: the compliance review that gets skipped, and the second proofreader who is really you at 11pm.

automaticACTIVE

The Scanner

Banned terms, policy tripwires, unresolved slots, raw hex, house style. It does not have opinions. It has an exit code.

TRYruns on every draft, before anything leaves

Will never: Pass anything on judgment.

/proofACTIVE

The Proofer

One piece, three fresh-context lenses: rules, facts, voice. Returns pass, flag or fail with line numbers. It cannot be talked into being the approver.

TRY/proof 05 Work Zone/review-queue/retention/welcome-letter-DRAFT.md

Will never: Approve, or rewrite the piece. It reports and a human decides.

/qaACTIVE

The Fleet Marshal

The same three lenses across the whole review queue at once, counting returns before reading them, then filing the ranked report.

TRY/qa

Will never: File a report with a missing verdict as complete.

/verifyACTIVE

The Authority Officer

Traces every number and every claim in a piece back to its authority file.

TRY/verify the partner deck

Will never: Accept a figure whose source is somebody’s recollection.

/auditBUILT

The Register Auditor

Diffs every status surface against every other and returns the contradictions with file and line on both sides.

TRY/audit

Will never: Reconcile by trusting whichever surface is newest.

/acceptACTIVE

The Examiner

Probes the brain cold, including refusal traps it is supposed to fail to answer.

TRY/accept run the acceptance test

Will never: Judge on self-report rather than on behavior.

/incidentBUILDABLE

The Incident Officer

The loop that closes after the mistake: what went out, who it reached, what gets corrected, and the new filed fact plus rule so the scanner catches that whole class forever after.

BUILDABLE: The honest gap in this roster. Today the loop ends at approval and never at outcome.

TRY/incident a wrong figure reached a partner

Will never: Close an incident without a rule that prevents the recurrence.

/escalateBLOCKED

The Escalation Officer

Routes money, legal and external-send decisions to the named human who cannot be bypassed.

BLOCKED: Who holds which approval is itself an open question in the register. It has to be answered before this can exist.

TRY/escalate this needs a signature

Will never: Be the bypass.

D9 · Engineering and Delivery Runs the machinery: builds, regenerates, certifies, promotes, deploys. Marked commands are Jordan*’s. Removes the need for: the developer you have to book to change one sentence.

/publishACTIVE

The Sitekeeper

Operator command. The whole ship-it pipeline as one careful sequence: rebuild, audit, deploy, sync, push. A failing gate kills the deploy.

TRY/publish

Will never: Deploy past a red gate.

/promoteACTIVE

The Promoter

Operator command. Moves an approved piece into the official library the safe way: copy, archive the old one, log the change.

TRY/promote the welcome letter, Avery* approved it this morning

Will never: Promote without a named human approval.

/regenACTIVE

The Generator

Regenerates every derived surface: directories, indexes, twins, constitution mirrors.

TRY/regen

Will never: Let a generated file be hand-edited.

/workflowACTIVE

The Fleet Commander

Runs wide jobs as coordinated fleets with independent verification wired in, one worker per file and a fresh-eyes checker on every return.

TRY/workflow rebrand the collateral library

Will never: Merge results it has not counted.

/batchBUILT

The Migration Lead

Sequences the collateral migration, one reviewed batch at a time, with Jordan* reading the batch report before the next one starts.

TRY/batch start the migration

Will never: Start batch two before a human approved batch one.

/accessBUILDABLE

The Access Warden

Owns permissions and what any connector may reach. It has to exist before any integration does.

TRY/access what can the brain reach today?

Will never: Grant an integration a scope nobody asked for.

The setup register · 57 questions tracked

Team Questions

Every question the intake asks, on one answerable page. The brain never guesses: anything unanswered stays here, numbered, with a destination — and in production, answers typed here file themselves into the right document.

DEMO. This is what the register looks like in production. The cards are typeable so you can feel the flow — here, Send simply marks the card. Each question shows the exact file its answer lands in.

Section 1 · Identity → the constitution · rules · facts

1.1→ {{company_legal_name}} + the house-style name rule

Legal business name (exact)

1.2→ Rules/house-style.md

The name rule: how it must always appear — and never appear

1.3→ CLAUDE.md §1 · positioning

One-liner: what you do, for whom, in one sentence

1.4→ Facts/about-company/company-overview.md

The 30-second elevator paragraph

1.5→ hard rules · positioning

What you do NOT do / never want to be mistaken for

1.6→ guardrails · authority files

Industry + regulated flags

1.7→ Facts · service-area rules

Geography served

1.8→ reference-index/published-pages.md

Web presence: domains, socials, live pages

Section 2 · People → governance · people-and-org

2.1→ the approval roster

Owner / final approver

2.2→ the approval roster

Brain operator — runs the AI day to day

2.3→ governance §8

Additional approvers + what each may approve

2.4→ reference-index/team-directory.md

Team roster: name · role · email · headshot

2.5→ guardrails

The money/legal/external-send gate: who can never be bypassed

Section 3 · Audiences → target-audience files

3.1→ Rules/target-audiences/

Audience 1: who · wants · fears · words that work · the action

3.2→ Rules/target-audiences/

Audience 2, same shape

3.3→ positioning

Which one is the hero audience (exactly one)

Section 4 · Voice → the voice trio + tones

4.1→ voice.md

Three to five voice adjectives

4.2→ writing-tones/

Default tone

4.3→ approved-phrases.md

Approved signature phrases — things you actually say

4.4→ banned-phrases.md

Banned words + phrases

4.5→ founder-voice reverse-engineering

Three writing samples you love

4.6→ house-style.md

Mechanics: emoji · exclamations · humor

Section 5 · Hard rules → guardrails

5.1→ guardrails.md

“This AI must NEVER…” — say, do, claim

5.2→ the pricing authority file

Pricing publishing policy

5.3→ claims-and-credentials.md

Claims we CAN prove vs claims we never make

5.4→ voice rules

Competitor mention policy

5.5→ guardrails

Anything sensitive the AI must handle carefully

5.6→ house-style.md

House mechanics: dates · phones · dashes

Section 6 · Facts → the fact layer

6.1→ Facts/services/ + pricing.md

Services/products: description · who for · price · flagship?

6.2→ the pricing AUTHORITY file

Pricing detail + what's quotable

6.3→ claims authority

Credentials + awards (verifiable only)

6.4→ company-overview.md

Founding story · history · key dates

6.5→ claims authority

Stats you can cite, each with a source

6.6→ Facts/faqs/

Top 10 client questions + your best answers

6.7→ never used without permission

Testimonials + permission status per quote

6.8→ Facts/source-docs/

Document uploads: decks · price lists · SOPs

Section 7 · Market → competitors · positioning

7.1→ verified by live search at build

Top 3–5 competitors + your one-line take

7.2→ location pages

Service area detail, ranked

7.3→ positioning

Why you win: 2–4 pillars

7.4→ SEO module

SEO ambition + target searches

7.5→ channel strategy

Where clients actually come from today

Section 8 · Brand → the Brand OS · stage 02

8.1→ routes the Brand OS build

Brand maturity: full system · logo + colors · from scratch

8.2→ Brand OS inputs

Uploads: logos · fonts · guidelines · collateral you love

8.3→ the token table

Known color codes

8.4→ direction briefs

Direction words + brands you admire

8.5→ the Exclusions Register

The visual never-list

Section 9 · Collateral menu → skills · templates

9.1→ one skill per pick, each on a gold standard

Pick from the catalog: print · web · email · decks · social · docs

9.2→ built first, become the reference standards

The first 3 pieces you need on day one

Section 10 · Modules & infrastructure → optional systems

10.1→ website module

Website: keep · rebuild in brain · none

10.2→ content module

Content/SEO engine?

10.3→ account-kit module

Partner/account kits + the account list

10.4→ work-zone variant

Team size: solo · 2–3 · team

10.5→ the scribe loop

Meeting/transcript filing?

10.6→ integration checklist (credentials never collected)

Accounts in play: repos · hosting · CRM · email sender

Section 11 · History → memory · the registers

11.1→ Facts/decisions/

Decisions already locked — things the AI must not reopen

11.2→ seeded into this register, numbered

Known unknowns

11.3→ ROADMAP.md

In-flight work: campaigns, launches, deadlines

What’s locked

Decisions Log

Every decision leadership has made, dated and sourced. Superseded rows stay, marked, because history matters. All rows below are demo examples.

DEMO. An example of what the decisions log looks like in production: every locked call, dated, sourced, and superseded rows kept for history. These rows mirror the implementation plan and the setup register.
WhenDecisionApproverStatus
Week 1New masterbrand selected + trademark filed; name locks at brand-lock; ™ until registrationDana* + counselLOCKED
Week 1Existing product lines keep their names through the transition windowLeadershipLOCKED
Week 1Public posture: one brand, told through customer continuity firstLeadershipLOCKED
Week 2Brand sprint scope: directions → leadership top-2 → one revision roundAvery*ACTIVE
Week 2New brain built fresh for Custom LLM (not a legacy-brain override); rules carried, branding rebuiltJordan* + Avery*ACTIVE
Week 2Atlas adopted as the identity; red retired outright; True North licensed as the one accent, and never as type on whiteAvery* + Jordan*ACTIVE
Week 2The six colour names locked into the Brand OS (Midnight Green, True North, Paper White, Landfall, Meridian, Surface Steel), each with the ratio it was measured atLeadershipLOCKED
Week 3Written retention terms adopted (rate lock + service guarantee); verbal figures supersededLeadershipLOCKED
Week 3Launch cascade locked: T-7 sign-off · T-4 manager briefings · go/no-go · launch day hour-by-hourLaunch planLOCKED
Week 3PR agency shortlist to three; selection within the weekCommsACTIVE

Open items live on the Q register — nothing here ships while its question is open.

Every change, dated

Changelog

The mandatory record: every change to the brain, what it was, why it happened, and who made it. Newest first. Entries below are demo examples of the real format.

DEMO. An example of the mandatory record in production — the entries below tell this demo build’s own story in the real format. Nothing ships without a line here.
Access · connection · your AI · the rules

How To Use the Brain

Everything a teammate needs on one page: how access works, how to connect the files and your AI, how to use the agents and skills, which model and effort to pick, and the habits that keep the brain honest. This demo shows Claude — the same pattern works with any capable AI.

Demo note. Everything on this page is demo content, and names marked * are demo placeholders, not real people.

What this is

The company brain is the team’s shared system: one library that holds what is true, how we sound, and what is approved, plus an AI that works inside it like a colleague. It lives in three connected places that always agree: the file library (plain files and folders — host them wherever the team already lives: SharePoint, OneDrive, Google Drive, Dropbox, Box, or a GitHub repo), the team site (the readable, browsable version of everything), and the AI (connected to the folder). You save work into it, the brain checks and files it, and the site updates so everyone sees the same truth.

This one document replaces the usual stack of guides. If something is not answered here, ask on the Team Questions page and the answer becomes part of the brain.

Access levels

Access is deliberately simple: everyone can read everything, you can only write in your own space, and one admin runs the machinery.

LevelWhoCan seeCan edit
AdminJordan* (operator)EverythingEverything, site settings, memberships, promotions into the brain
ApproverAvery*, Morgan*EverythingTheir own Work Zone, plus approve/return anything in any Send for Approval folder
Team memberSami*, Frankie*, Dana*, Gale*, Marley*Everything (view only)ONLY their own Work Zone folder

Folder-level permissions enforce this on whichever platform hosts the library. Until every level is wired, the working rule is the same: edit only inside your own Work Zone; everything else is read-only by convention.

Getting connected

  1. The brain is just files and folders, so it lives happily on SharePoint, OneDrive, Google Drive, Dropbox, Box, or GitHub — whichever the company already uses. This demo assumes a synced cloud folder.
  2. Sync it locally with that platform’s own tool (Sync / “Available offline” / clone), and pin it so it is always on your device.
  3. That folder IS the brain. Anything you save inside it is in the library seconds later; anything a teammate saves appears for you the same way. No uploading, no emailing files.

Where things live

You wantGo to
Orientation, this guide00 Start Here/
The AI’s knowledge: facts, rules, approved library01 The Brain/
Brand system and design assets02 Brand/
Source documents, transcripts, price files03 Knowledge and Sources/
Plans, the strategic plans04 Plans/
YOUR working space05 Work Zone/<your name>/
The browsable everythingThe team site: Assets tab (every file, color-coded), Agents, Skills, Plan, the Batch Board

Your Work Zone

Five folders, and that is the entire system: 00 My Space (yours, no rules) · 01 Active Projects · 02 Send for Approval · 03 Save for Later · 04 Archived. Drop finished work into Send for Approval; Jordan*, Avery*, or Morgan* reviews it; approved work gets built into the brain and appears on the site, and anything needing changes comes back to your Active Projects with a note. Nothing is ever deleted, it retires to Archived.

The rules that keep the brain honest

  1. Launch-gated above all — drafts and launch material stay internal until the launch date; the scanner enforces it.
  2. Nothing invented — the brain never guesses; unknowns become numbered questions on the Team Questions page. You should work the same way.
  3. Approved is sacred — never edit anything in the Approved library directly; changes go through the Work Zone and re-approval.
  4. Markdown is canonical — every document’s .md is the source of truth; the styled page beside it regenerates automatically. Edit the .md, never the .html.
  5. Ask, don’t assume — the Questions page routes every open item to its owner, and answers take effect the day they arrive.

Who to ask

Jordan Vale* — operator, AI owner, access, anything technical. Avery Kellen* — marketing lead, approvals, priorities. Everything else: the Team Questions page on the site.


The AI side of the brain. Connected to the library, the AI answers from filed facts with citations, drafts in the house voice, checks work before approval, and runs the machinery. This demo runs on Claude; the library also generates AGENTS.md, a universal mirror of the constitution, so ChatGPT, Gemini, Copilot, or any other tool the team prefers inherits the identical rules.

Connecting your AI to the brain

The three surfaces, when to use which (this demo’s stack)

SurfaceWhat it isUse it for
ChatThe browser conversationQuick thinking, wording help, anything that needs no files
CoworkThe AI working inside a folder with youDay-to-day brain work: drafting, reviewing, filing, “find me X”
CodeThe terminal operator surfaceAgents, skills, batches, site builds — the machinery. Team members mostly will not need it; Jordan* lives here

The agents (in a Code session at the brain root)

Type the slash command, add your request, press enter. Each specialist runs with fresh eyes and reports back. The full roster with examples is the site’s Agents tab, and every specialist ships real work — browse example outputs ↗. The daily four:

The skills

Skills are packaged abilities: you do not type commands, you just ask for the thing (“scan this file”, “build the deck in the Kai* style”, “clean up the BioSummit folder”) and the brain routes to the right skill. The full library with what each does is the site’s Skills tab. Fourteen generation skills are built and switch on automatically at brand-lock.

Models: which brain to rent

The table shows Claude’s lineup, the demo’s stack — pick the equivalent tier in whichever AI the team runs.

ModelCharacterUse for
HaikuFastest, lightestQuick lookups, simple summaries, high-volume small tasks
SonnetThe balanced daily driverEveryday drafting and questions when speed matters
OpusDeep, careful, thoroughThe standard for brain work — anything touching facts, approvals, or deliverables. All brain agents are pinned to Opus
FableThe frontier tier above OpusThe hardest builds and largest campaigns; use deliberately, it costs the most

Effort levels

Some surfaces let you set how hard the model thinks: Low answers fast with minimal deliberation (lookups, formatting); Medium is the default balance; High reasons deeply before answering (analysis, plans, anything with edge cases). Rule of thumb: raise the effort when being wrong is expensive, lower it when speed matters more than polish.

The habits that make the AI useful here

  1. Give it the file, not a description of the file. A connected AI reads the real thing.
  2. Ask for sources. A brain answer should cite where the fact lives; if it cannot, the answer belongs on the Questions page.
  3. Let it check your work/proof before anything goes up the chain costs one minute and catches what tired eyes miss.
  4. Save into your Work Zone and let the approval flow move it; never ask the AI to write into the Approved library directly (it will refuse anyway).
  5. When the AI says UNKNOWN, that is the system working. File the question rather than pushing for a guess.
The Process · from zero to a working brain

How one of these gets built

Four stages, about two weeks, one system at the end: a governed library your AI reads before it writes a word — plus the site, the agents, the skills, and a team that knows how to use it.

Why AI adoption stalls. Everyone has access to AI; almost no team can get on-brand, consistent, approved work out of it. Ten people prompting ten ways produce ten different companies on paper. The fix isn’t a better prompt — it’s a system the AI is required to read: what’s true, how you sound, what’s approved, and rules it cannot break.

What you end up with

The library

One governed source of truth: facts with authority files, voice rules, approved templates, and a scanner that gates every deliverable.

The machinery

A team of specialist agents and packaged skills — drafting, checking, filing, and wide jobs run as verified fleets.

The surfaces

This site pattern: browsable assets, plans, questions, decisions, changelog — the whole brain visible to the whole team.

The habits

A trained team, an approval flow that moves work by moving folders, and a register that turns unknowns into answers.

Want one built?

This whole system — the intake, the Brand OS, the brain, the training — is something I build for companies as an engagement. Email me and we’ll set up a time to talk through yours.

contact@djvonfrank.com
The Process · stage 01

The Intake Form

Everything needed to set up the brain, in one pass. Every field maps to a real file in the library — and any field left blank ships as a numbered question in the new brain’s register, because the brain never guesses.

Demo copy. The fields below are typeable so you can feel the flow, but nothing saves or sends. In a real engagement this form can be completed incrementally — unanswered fields become tracked questions, not blockers.

Section 1 · Identity

→ constitution · rules · facts/about-company
→ {{company_legal_name}} + the house-style name rule
→ Rules/house-style.md
→ CLAUDE.md §1 · positioning
→ Facts/about-company/company-overview.md
→ hard rules · positioning
→ guardrails · authority files
→ Facts · service-area guard rules
→ Facts/reference-index/published-pages.md

Section 2 · People

→ governance · facts/people-and-org
→ the approval roster
→ the approval roster
→ governance §8
→ reference-index/team-directory.md
→ guardrails

Section 3 · Audiences

→ target-audience files (2–4, one is the hero)
→ Rules/target-audiences/
→ Rules/target-audiences/
→ positioning

Section 4 · Voice

→ the voice trio + tones
→ voice.md
→ writing-tones/ (reassuring-advisor · compliance-clear · plain-teacher · peer-professional · executive-brief · founder-voice · high-energy-promo)
→ approved-phrases.md
→ banned-phrases.md
→ founder-voice reverse-engineering
→ house-style.md

Section 5 · Hard rules

→ guardrails, the ten rules
→ guardrails.md
→ pricing authority file
→ claims-and-credentials.md — AUTHORITY FILE
→ voice rules
→ guardrails
→ house-style.md

Section 6 · Facts

→ the fact layer (every number needs one home)
→ Facts/services/ + pricing.md
→ the pricing AUTHORITY file
→ claims authority
→ company-overview.md
→ claims authority
→ Facts/faqs/
→ never used without permission
→ Facts/source-docs/ — preserved verbatim

Section 7 · Market

→ competitors · positioning · modules
→ verified by live search at build
→ location pages
→ positioning
→ SEO module trigger
→ referral · search · social · partners · outbound

Section 8 · Brand

→ the Brand OS (stage 02)
→ routes to the Brand OS build
→ Brand OS inputs
→ token table
→ direction briefs
→ Exclusions Register

Section 9 · Collateral menu

→ skills · templates · the routing table
→ one generation skill per pick, each built on a gold standard
→ become the reference gold standards, built first

Section 10 · Modules & infrastructure

→ optional systems
→ website module
→ content module
→ account-kit module
→ work-zone variant
→ the scribe loop
→ integration checklist

Section 11 · History

→ memory · the registers
→ Facts/decisions/
→ seeded into the questions register, numbered
→ ROADMAP.md

What submitting produces

the output contract

A complete Intake Packet: every answer keyed to its destination file · every unanswered field numbered into the questions register · an upload manifest with a home for every document · a completion percentage. The packet plus the signed-off Brand OS is everything the two-week build needs.

Want one built?

This whole system — the intake, the Brand OS, the brain, the training — is something I build for companies as an engagement. Email me and we’ll set up a time to talk through yours.

contact@djvonfrank.com
The Process · stage 02

The Brand OS

Creating this is the first build step of the LLM setup — and it’s the step most AI rollouts skip, which is why they stall.

The bottleneck, named. AI adoption hasn’t stalled because models can’t write — it’s stalled because teams can’t get on-brand collateral in a consistent format out of them. Every person prompts differently; every output looks different; nothing survives brand review. The Brand OS solves exactly that: one token table and twenty ready treatments that every future output inherits. Change the tokens, and the entire collateral system re-renders in the new brand.

The master, in full

This is the actual master template every engagement starts from — five chapters (Token Schema · UI Pattern Skeleton · Collateral Layer · Swap Protocol · Exclusions Register), 20 collateral treatments, 395 fill-slots, zero hardcoded colors, rendered live below on the neutral “Placeholder & Co.” identity.

T01 LetterheadT02 One-pagerT03 ProposalT04 Leave-behindT05 Card + signature + QRT06 OG imageT07 Email chromeT08 Deck themeT09 Social framesT10 Document chromeT11 Price sheetT12 Welcome flyerT13 Checklist guideT14 Intake formT15 Microsite canonT16 Landing pageT17 Executive summary pairT18 Blog treatmentT19 Trifold + catalogT20 Mailer + certificate

How the swap works

01 CLONE

Start from the master

The master is brand-agnostic by construction: every color, font, and name lives in one token table. Nothing is hardcoded anywhere else.

02 SWAP

Run the Swap Protocol

The client’s palette, type, marks, and name replace the tokens. All twenty treatments re-render in the new brand in one pass.

03 LOCK

Audit, then sign off

Contrast is measured, exclusions are registered, leadership signs. The treatments become the gold standards every generation skill copies from.

Want one built?

This whole system — the intake, the Brand OS, the brain, the training — is something I build for companies as an engagement. Email me and we’ll set up a time to talk through yours.

contact@djvonfrank.com
The Process · stages 03 + 04

The Build, then the Team

About two weeks from intake packet to a working brain — then the part most rollouts skip: making sure the team actually uses it.

Week 1 — the skeleton and the truth

DaysShips
1–2Charter · the six-zone tree · the registers · the constitution v1 with the ten hard rules and the routing table
3–4Fact ingestion from the intake packet · the two authority files (pricing, claims) · the voice trio and audiences
5The scanner: banned terms, policy tripwires, slot checks — wired into every handover

Week 2 — the system that does work

DaysShips
6–7Work-zone lifecycle (position is status) · Brand OS swap locked · tokens wired
8–9Generation skills on gold-standard references · the agent team, staged on real gates
10The QA fleet’s first live run · the acceptance test · team walkthrough

The full dated version lives in the project plan, and the gates that pace everything after (brand-lock, launch, phase 2) are on the calendar.

Stage 04 · the team, onboarded

Session 1 · The tour

Read-only walkthrough: the site, the zones, where truth lives, how position-is-status works. Everyone leaves able to find anything.

Session 2 · First drafts

Each person ships one real piece through the full loop: ask → draft → proof → Send for Approval → approved. The flow becomes muscle memory.

Session 3 · The machinery

Agents and skills for the daily drivers; the questions register as the team’s habit; operators learn the deeper surface.

The artifacts

The How To page and Guides & Tools stay with the team — onboarding that outlives the onboarder.

Want one built?

This whole system — the intake, the Brand OS, the brain, the training — is something I build for companies as an engagement. Email me and we’ll set up a time to talk through yours.

contact@djvonfrank.com
Playbooks · the meta-story

How This Demo Was Built

This site is a working rebuild of a real production system — de-branded page by page with the same discipline the brain itself teaches.

The short version

  1. A real brain, in production. The original runs a live engagement: same architecture, same agents, same gates — with a real company’s facts inside.
  2. A full duplicate, kept local. The copy never touched a public URL while it still held client material; it ran on a local server the whole way.
  3. De-branded page by page. Every company, partner, and person was mapped to a demo replacement — same shape, zero real data. Names on prose carry an asterisk; the fictional client is Apex Instruments*, mid-rebrand, its new brand still an open {{slot}}.
  4. The deep pass. 600+ library files were content-scrubbed in place; anything a rename couldn’t make safe — transcripts, price data, source binaries — was neutralized and unlinked, marked Not in demo right in the tree.
  5. Gates, not vibes. Every pass ended with scrub-gate greps across every file, link crawls, and console checks. A clean-looking page that fails a gate doesn’t ship — the demo was built under the same rule it demonstrates.

What’s real vs what’s demo

RealDemo
The architecture, the code, the tree shape, the operating loop, the agents-and-skills design, the QA-fleet methodEvery company, person, partner, number, price, date, and logo

The tooling

Built and de-branded in Claude Code sessions — the same operator surface the brain teaches — with scripted map passes for the sweeps and hand editing for everything with judgment in it. The full change log is on the changelog, demo-formatted like everything else.

Want one built?

This whole system — the intake, the Brand OS, the brain, the training — is something I build for companies as an engagement. Email me and we’ll set up a time to talk through yours.

contact@djvonfrank.com