# Marketing OS
Every's AI-native marketing operating system. Four workflows connect strategy, market
execution, words, and visuals. Useful corrections improve the next comparable piece of work.
Marketing creates demand and cultural gravity. Growth converts that demand. Marketing OS
connects the two through a measurable operating loop while protecting creative ambition
and brand coherence.
## How It Works
Ask Marketing OS for the outcome, or invoke one workflow directly:
| Workflow | Job | Entry |
|---|---|---|
| Compound Strategy | Decide what to do, for whom, and why | [Start](workflows/compound-strategy/SKILL.md) |
| Compound GTM | Bring approved strategy to market | [Start](workflows/compound-gtm/SKILL.md) |
| Compound Copywriting | Write, review, and improve the words | [Start](workflows/compound-copywriting/SKILL.md) |
| Compound Art Direction | Direct and produce the visual system | [Start](workflows/compound-art-direction/SKILL.md) |
These are independent entry points, not four mandatory stages. Marketing OS coordinates
them for larger projects using one approved brief or decision record. Each uses the same
[learning protocol](foundation/marketing-os/references/compound-workflows.md): make,
review, capture a supported correction, apply it to fresh work, and check the result.
The theoretical foundation, brand context, and specialist guidance remain intact.
`strategy/compound-marketing` stays available as the legacy end-to-end coordinator;
existing specialist paths and saved run records continue to work.
## Install or share
Install the **whole repository as one skill folder** named `every-marketing-os` in your
host's skill directory. The root [SKILL.md](SKILL.md) routes into the four workflows.
Do not copy an individual workflow folder: it depends on shared references elsewhere
in this repository. The repository root, not the shell's current directory, anchors
repository-relative paths.
For an existing setup that prefers four separately discoverable workflows, keep the
full checkout and use thin adapters pointing to its four public entries. Choose either
that setup or the umbrella entry to avoid duplicate discovery. Supporting `SKILL.md`
files are routed references, not separately installable packages.
Python 3.10+ and Git are needed only for release checks and packaging, not for reading
the skill. From the repository root:
```sh
python3 scripts/validate_release.py
python3 scripts/validate_release.py --zip dist/every-marketing-os.zip
```
The ZIP includes the complete release candidate and excludes ignored private notes,
local integration settings, and generated output. Untracked, nonignored additions are
included, so review the candidate before publishing. Checks cover local references,
JSON, public entry metadata, and selected privacy patterns; they do not certify the
absence of sensitive material or prove better marketing outcomes.
Slack, Notion, Google Drive, Kit, and specialist reviewers are optional host capabilities,
not bundled connections. Without them, use supplied sources, identify missing evidence,
and continue independent work. Product scaffolds remain explicitly unfinished.
The repository currently has no declared license. Confirm licensing with the repository
owner before redistribution through a skill library.
Routing regression scenarios live in [workflows/evals/routing-cases.json](workflows/evals/routing-cases.json).
They complement the foundation smoke cases; synthetic checks do not establish market results.
Every specialist skill still loads `foundation/marketing-os` as its root. Research,
audiences, positioning, voice, craft, launches, and measurement remain available when
the active stage requires them.
`workflows/` contains the four public entries. The supporting repository map is:
1. **foundation/** — The Marketing OS, the operating system underlying all skills
2. **brand-voice/** — Voice guides for Every's master brand and each sub-brand (Cora, Spiral, Monologue, Plus One, Proof, Sparkle)
3. **positioning/** — Positioning frameworks for Every master and each sub-brand
4. **strategy/** — Strategy specialists, full-brand workflow, and the shared run coordinator
5. **craft/** — Execution-level skills for art direction, copywriting, Compound Copywriting, editing, naming, and channel-specific output (launch emails, LinkedIn posts, X posts, website copy, press comms)
6. **launches/** — Orchestration skills for three launch tiers (improvement, feature, new product) that load strategy and all channel craft skills
7. **marketing-science/** — Research, archetyping, brand equity, and measurement skills that inform positioning, programs, and decisions
8. **marketing/** — Integrated planning workflows that turn strategy, sources, and campaign decisions into executable GTM artifacts
## Dependency model
Every skill loads `foundation/marketing-os` as its root dependency. Additional dependencies cascade by category — positioning skills pull in marketing-science, craft skills pull in brand-voice and positioning, launch skills orchestrate strategy and all five channel craft skills.
See [ARCHITECTURE.md](ARCHITECTURE.md) for the full dependency graph.
## Taxonomy
The four public workflows use these eight supporting families:
1. **foundation/** — Every's marketing doctrine, canon, operating frameworks,
collaboration rules, and native-document editing protocol
2. **brand-voice/** — Voice systems for Every and its products
3. **positioning/** — Durable, ruled master-brand and product positioning
4. **strategy/** — Compound Marketing, Compound Brand, messaging architecture,
measurable program briefs, and executive one-pagers
5. **craft/** — Verbal identity, art direction, copywriting, Compound Copywriting,
editing, naming, PR,
social, email, web, and long-form execution
6. **launches/** — The canonical launch process, launch calendar, brief, GTM strategy,
claims clearance, wireframing, email flows, and three launch tiers
7. **marketing-science/** — Research, audience archetypes, brand equity, messaging
evidence, attribution, Fame Score, cohort economics, and results compounding
8. **marketing/** — Integrated GTM planning that turns approved strategy, sources, and
campaign decisions into executable artifacts
Optional historical cases live in `docs/cases/`. Skills never load them by default.
## What Belongs in the OS
Marketing OS stores reusable methods: how to turn context into strategy, execute a
launch, build a GTM plan or recurring program, measure the result, and compound the
learning.
It does not store active roadmaps, launch briefs, campaign concepts, current rosters,
temporary claims, prices, budgets, embargoes, or project calendars. Durable positioning
and voice may live in their product modules only after they have been ruled as canon.
Project work can change the OS when Douglas explicitly makes the learning a standard,
the pattern succeeds across multiple relevant projects, or unusually strong evidence
supports a bounded rule. Otherwise the learning stays with the project or in an optional
case.
## Current State
The four-workflow interface organizes existing capabilities; it does not establish new
performance claims or promote unfinished specialist skills to production-ready status.
Ready for structured use:
- Every master positioning, brand voice, audience system, and brand architecture
- Evidence-led research, competitive auditing, and positioning matrices
- Compound Brand strategy through verbal identity and art direction
- Messaging architecture, proof standards, and evidence banks
- Canonical launch planning, GTM strategy, claims clearance, and channel production
- Program briefs with owners, costs, subscriber paths, and investment gates
- PR and Instagram planning tied to authority and subscriber acquisition
- Executive one-pagers for 10- to 15-minute leadership reviews
- Marketing measurement across sourced free subscribers, cohorts, CAC, LTV:CAC, Fame
Score, authority, and Consulting demand
- Compounding from Douglas's edits and from campaign results without importing project
details into unrelated work
- Optional marketing pods that run research, operations, production, QA, and measurement
as bounded agent roles while Douglas retains strategic and creative control
- A Compound Marketing golden path with decision records, run records, cross-stage
fidelity checks, and an explicit proof standard
- Compound Copywriting beta that interviews incomplete briefs, routes by persuasion job,
selects one copywriter review lens, turns human edits into bounded project instructions,
and tests transfer while preserving human creative control
Still in development:
- Product-specific positioning and voice where source documents have not been ruled
- Several execution-channel skills that remain explicit scaffolding
- More forward tests and comparative evaluations
- Live Compound Copywriting baselines and transfer evidence across comparable copy tasks
- Live data integrations for repeatable quarterly reporting
- A prospective comparable second run and independent teammate rerun required for a
publishable compounding claim
The Brand Book and Thesis GTM cases are recorded retrospective baselines. They pass the
structural record checks and intentionally report `proof_ready: false`; neither recorded
enough prospective evidence to prove compounding.
Files marked `## To do` are scaffolding, not authoritative product or channel guidance.
## Changelog
### Oct. 2, 2026: Four public workflows
- Organized entry points into Compound Strategy, GTM, Copywriting, and Art Direction
- Kept theory, brand context, specialist implementations, and legacy paths intact
- Unified discipline-specific correction and transfer checks under one shared protocol
- Preserved existing run records and the distinction between capture and proven improvement
### Aug. 21, 2026: One GTM and one compounding system
- Consolidated flagship and standard GTM into one canonical `marketing/gtm` entry point
- Preserved the full customer, offer, campaign, proof, conversion, event, and measurement
doctrine as routed GTM references
- Added explicit flagship, standard, scoped-revision, drumbeat, and social operating modes
- Unified Compound Marketing, Compound Copywriting, and optional pods around the same
decision records, knowledge boundaries, and proof standard
- Removed mandatory intake questions when the request and authorized sources already
settle the governing brief
### Aug. 19, 2026: Compound Marketing V1
- Added the governing Compound Marketing strategy and canonical golden path
- Added one entry skill for strategy, recurring programs, GTM, and learning closeout
- Added shared decision and run records that preserve human rulings across stages
- Added deterministic validation for decision fidelity and comparable-run proof
- Added honest Brand Book and Thesis GTM baselines without upgrading them into claims
- Kept design, asset production, and agent-count claims outside V1
### Aug. 19, 2026: Compound Copywriting
- Added the Write → Judge → Learn → Test → Repeat copy-production loop
- Separated immediate project learning from gated durable doctrine
- Added revision-burden, protected-language, quality, hard-failure, and transfer metrics
- Added six regression cases for human authorship, transfer, overfitting, and false
measurement claims
### July 24, 2026: Reusable Operating Layer
- Added Marketing-sourced free subscribers, attribution, cohort quality, CAC, LTV:CAC,
Fame Score, authority, and Consulting demand measurement
- Added program briefs that preserve the creative idea while defining ownership, cost,
instrumentation, and scale, revise, or stop gates
- Added reusable PR and Instagram planning methods
- Added executive summaries for 10- to 15-minute leadership reviews
- Added native-document editing and compounding from Douglas's edits
- Added results compounding so completed programs create dated learning without turning
one campaign into permanent doctrine
- Added knowledge-boundary rules that separate reusable doctrine, durable brand canon,
and optional historical cases
- Removed active project plans, launch evidence, rosters, offers, and dated campaign
instructions from default skill paths
### July 18–21, 2026: Brand and Launch Systems
- Built the Every master brand voice
- Added the canonical marketing launch workflow and GTM plan
- Built Compound Brand strategy through art direction
- Strengthened GTM decision, conversion, and activation rules
### July 6–13, 2026: Execution Foundation
- Populated research, archetyping, positioning, and brand-equity systems
- Added launch architecture, claims clearance, PR checks, and email-flow guidance
- Added messaging-document discipline, copywriting rules, and evidence-bank standards
- Began compounding approved human edits into durable operating guidance
See the [full commit history](https://github.com/EveryInc/Marketing-OS/commits/main) for
line-level changes.