The operating system for WeLaunch — a centralized hub for brand, clients, content, operations, and knowledge. A private git repository is the document layer (
welaunchllc/welaunch-os, checked out at~/Dev/welaunch-os); Google Drive holds only the Inbox and large media. Each connected tool owns its domain. Agents and humans navigate the same structure.
Last updated: 2026-09-07 (git cutover)
How this works
WeLaunch OS is not a single platform. It's a document architecture in a git repository connected to a broader ecosystem of tools, each owning what it's best at. The repo stores strategy, drafts, frameworks, reference material, scripts and small assets; every session and every automation lands its writes as a commit, so history, diffs and conflict detection come for free. Google Drive (WeLaunch Drive) keeps the Atlas Inbox, client-shared folders and anything over 1MB. HubSpot publishes, sells, and manages relationships. Figma designs. Obsidian connects ideas. Composio orchestrates agents across all of it. Atlas is the AI assistant that operates across the entire system.
The core rule: Every piece of information has one home. If it lives in HubSpot, don't duplicate it in the repo. If it lives in the repo, don't recreate it in Drive. Point to things, don't copy them.
Folder structure
WeLaunch OS/
├── 00 Brand
│ ├── Brand Blueprint
│ ├── Logo + Mark
│ ├── Color + Typography
│ ├── Photography + Imagery
│ └── Social Assets
│
├── 01 Clients
│ └── [Client Name]/
│ ├── Client Index ← Start here (manifest file)
│ ├── Ideas — [Client] ← Idea backlog + project-staging format
│ ├── Systems — [Client] ← Registry of running systems/programs + versions
│ ├── Discovery + Onboarding
│ ├── Brand
│ ├── Strategy
│ ├── Content ← If applicable
│ ├── Communications
│ ├── Reporting
│ ├── HubSpot/ ← Client portal config (properties, pipelines, workflows)
│ ├── Projects/ ← One subfolder per initiative (+ Project Index w/ HubSpot ID)
│ └── Reference Library/ ← Reusable / reference copies (was Deliverables)
│
├── 02 Playbooks
│ ├── HubSpot Onboarding
│ ├── RevOps Audit
│ ├── Content Engine Build
│ └── Managed Growth Services
│
├── 03 Templates
│ ├── Proposal Template
│ ├── SOW Template
│ ├── Case Study Template
│ ├── Audit Report Template
│ ├── Monthly Report Template
│ ├── Email Sequences
│ └── Client Index — TEMPLATE
│
├── 04 Content
│ ├── Blog + Articles/
│ │ ├── RevOps Demystified (Pillar 1)
│ │ ├── HubSpot Unlocked (Pillar 2)
│ │ ├── The Full-Stack Perspective (Pillar 3)
│ │ └── Growth for the Real World (Pillar 4)
│ ├── LinkedIn
│ ├── Newsletter
│ ├── Lead Magnets
│ └── Content Calendar
│
├── 05 Operations
│ ├── SOPs
│ ├── Finance
│ ├── Legal
│ └── Strategy
│
├── 06 Knowledge Base
│ ├── HubSpot
│ ├── Industry + Competitive
│ ├── Frameworks + Models
│ ├── ICP + Personas
│ ├── Sales
│ ├── Sources ← Raw dated external input (feeds), not vetted knowledge
│ └── Webflow
│
├── 07 LaunchPad ← WeLaunch products — home of the live Client Portal
│
├── 08 Atlas
│ ├── Atlas Brief ← Personality + operating principles
│ ├── Decision Log ← Self-updating intake queue
│ ├── Memory ← Declarative + episodic memory (Sessions, Errors, Outputs)
│ ├── Identity ← Atlas's own mark, motion, email assets
│ ├── Automations ← Atlas Automation Register
│ └── Tools + Integrations ← Tool connection docs, API notes
│
└── 09 Personal ← Conner's individual life-admin (not business)
├── Health Insurance ← Coverage decisions + renewal notes
└── Taxes ← Estimated quarterly + personal tax
What each folder does
00 Brand
The identity system. Brand Blueprint, logos, color swatches, typography files, approved photography, and social media assets. This is the visual and messaging source of truth. Everything here informs how WeLaunch shows up externally.
01 Clients
One subfolder per client, each following a standardized template. Every client folder starts with a Client Index — a manifest file that links to the client's HubSpot portal, Figma project, tool stack, and engagement details. Contact and stakeholder details live in HubSpot CRM, not here.
The Brand subfolder is standard across all clients. It holds the client's brand assets, voice and tone guides, visual guidelines, and any logo or imagery references. The Content subfolder is included only for clients where WeLaunch is delivering ongoing content work (editorial calendars, blog drafts, pillar strategy). For audit-only or HubSpot-build engagements, it can be omitted.
02 Playbooks
Repeatable service delivery frameworks. Each playbook documents how WeLaunch delivers a specific service — the steps, the sequence, the decision points, the expected deliverables. Playbooks are prescriptive: they tell you (or Atlas) what to do. Tagged as Agent Rules.
03 Templates
Blank starting points for deliverables. Proposals, SOWs, case studies, reports, email sequences, and the Client Index template. Templates are the raw materials that get copied and populated for each engagement. Templates are never modified directly — always copy to the destination folder first, then fill in the copy. Tagged as Agent Inputs.
04 Content
WeLaunch's own thought leadership and marketing assets. This is not for client content work — that lives in the client's folder. Blog drafts are organized by content pillar to mirror the structure in HubSpot Content Hub. LinkedIn posts, newsletter issues, lead magnets, and the content calendar all live here.
05 Operations
Internal business management. SOPs, finance (P&L, expenses, tax docs), legal (MSAs, NDAs, contracts), and Strategy/ — WeLaunch's own positioning, market sizing, service-line design, and Ideas — WeLaunch.md, the parked-idea backlog for the business itself. Purely about running the business — AI and automation configuration lives in 08 Atlas.
Strategy moved here from the Knowledge Base on 2026-08-24: the KB is proven, reusable reference material, and WeLaunch's own unvalidated strategy work is neither (Conner).
06 Knowledge Base
Reference material that both humans and Atlas pull from. HubSpot platform documentation and workarounds, industry and competitive research, strategic frameworks, and ICP/persona profiles. Proven material only — WeLaunch's own in-progress strategy and parked ideas live in 05 Operations/Strategy/. This is the "context" layer — the information that informs decisions and recommendations. Tagged as Agent Context.
07 LaunchPad
WeLaunch's productized offerings. Home of the live WeLaunch Client Portal (portal.welaunch.us) — the secure per-client delivery product (Next.js + Supabase + Vercel, two-way HubSpot sync via n8n). Build history and portal docs live here. Future client-facing tools, dashboards, or productized services also live in this folder.
08 Atlas
Home for the AI assistant layer. The Atlas Brief defines personality and operating principles. The Decision Log captures decisions that need to be reflected in config files — run /update-atlas to process it. Atlas Automation Register documents all automated cron routines. Tools + Integrations holds tool connection documentation and API notes. Slash commands for Claude Code live in .claude/commands/ at the project root — that's a Claude Code technical requirement, not a Drive folder. This is the folder that grows as WeLaunch's AI capabilities expand.
09 Personal
Conner's individual life-admin, kept separate from the business OS. Health insurance and benefits, personal finance, and personal taxes. Routing test: if something would only ever matter to Conner as an individual (not WeLaunch the company), it lives here. Business finance and legal stay in 05 Operations. Stores decisions and references only, never secrets.
What lives outside the repo
The git repo is the document layer. These tools own everything else:
| Tool | What it owns |
|---|---|
| HubSpot CRM | Contacts, companies, deals, pipeline, activity timeline |
| HubSpot Marketing Pro | Email campaigns, automations, workflows, forms |
| HubSpot Content Pro | Published blog posts, website pages, landing pages |
| HubSpot Commerce Pro | Quotes, invoices, payment links |
| Figma | Design files, mockups, prototypes (one project per client + one for WeLaunch) |
| Composio | Agent orchestration and authenticated tool execution across WeLaunch's own accounts. No client portal is connected; client access runs on per-client Service Keys |
Google Drive (WeLaunch Drive) | Atlas Inbox drop zone, per-client Shared/ folders, media over 1MB, Google Docs, licensed asset library |
| GitHub Actions and claude.ai routines | Scheduled automations that read and write the repo (weekly review, meeting notes, HubSpot updates feed, task runner) |
Content flow
Content flows in one direction: the repo → HubSpot. Never the reverse.
- Drafting happens in the repo (04 Content), either by you or by Atlas
- Review happens in the repo — edits, feedback, approval
- Publishing happens in HubSpot — either manually (copy to Content Pro) or via Composio
| Content type | Drafts in repo | Publishes to |
|---|---|---|
| Blog posts | 04 Content / Blog + Articles / [Pillar] | Content Pro |
| Website pages | 04 Content (or 00 Brand for core pages) | Content Pro |
| Email campaigns | 04 Content / Newsletter (or 03 Templates) | Marketing Pro |
| LinkedIn posts | 04 Content / LinkedIn | LinkedIn (manual or scheduled) |
| Lead magnets | 04 Content / Lead Magnets | Content Pro (landing page + download) |
File routing
Every file has a home. Here's how to decide where it goes.
Client-specific work → client folder, always
| Output type | Destination |
|---|---|
| Reports, audits, QBRs | 01 Clients/[Client]/Reporting/ |
| Strategy docs, roadmaps | 01 Clients/[Client]/Strategy/ |
| Content drafts (for client) | 01 Clients/[Client]/Content/ |
| Project work (HubSpot builds, specs, workflow docs, designs) | 01 Clients/[Client]/Projects/[Project Name]/ |
| Reusable / reference copies of finished work | 01 Clients/[Client]/Reference Library/ |
| Meeting notes, key decisions | 01 Clients/[Client]/Communications/ |
| Intake notes, kickoff docs | 01 Clients/[Client]/Discovery + Onboarding/ |
WeLaunch content → by pillar and channel
Blog drafts go to 04 Content/Blog + Articles/[Pillar Name]/. LinkedIn, newsletter, and lead magnets go to their respective subfolders.
Reusable learnings → extract to the OS
When client work produces something reusable:
| Discovery | Extract to |
|---|---|
| HubSpot workaround or tip | 06 Knowledge Base/HubSpot/ |
| Reusable framework or model | 06 Knowledge Base/Frameworks + Models/ |
| Industry or competitive insight | 06 Knowledge Base/Industry + Competitive/ |
| Process gap or improvement | Update the playbook in 02 Playbooks/ |
The rule
Client folders accumulate specific outputs. The OS evolves from reusable learnings. When in doubt, save to the client folder first, then flag if something looks worth extracting.
Templates are never modified directly. Copy the template into the destination folder, rename with the client name and date, then populate the copy. The original in 03 Templates/ always stays blank.
Naming conventions
Consistent naming lets agents parse file purpose, client, and date without opening the file.
Files
[Client Name] — [Doc Type] — [YYYY-MM]
Examples:
Acme Corp — RevOps Audit — 2026-05Acme Corp — Monthly Report — 2026-04Acme Corp — Content Strategy — 2026-03
Client folders
Use the company name exactly as it appears in HubSpot. The same name is used for the repo client folder and the WeLaunch Drive/Clients/ media folder, so both link cleanly from the CRM record.
Number prefixes
Top-level folders use 00–09 prefixes to force sort order in any file browser. This keeps the structure consistent regardless of who's browsing — human or agent.
Agent roles
Folders are tagged by how Atlas interacts with them:
| Tag | Meaning | Folders |
|---|---|---|
| Agent rules | Frameworks and processes Atlas follows when delivering work | 02 Playbooks |
| Agent inputs | Blank templates Atlas populates with client-specific content | 03 Templates |
| Agent context | Reference material Atlas pulls from to inform decisions | 06 Knowledge Base |
An agent building a monthly report would: read the Managed Growth Services playbook (rules) → grab the Monthly Report Template (input) → pull relevant frameworks and ICP data (context) → draft the report into the client's Reporting folder (output).
Atlas
Atlas is WeLaunch's AI assistant — a single, unified teammate that operates across the entire OS. Not a panel of fake executives or a hierarchy of specialized agents. One assistant with full access to playbooks, templates, knowledge, and client context.
Atlas's configuration lives in 08 Atlas/:
- Atlas Brief — Personality, voice, operating principles
- Decision Log — Self-updating intake queue (run
/update-atlasto process) - Atlas Automation Register — All active cron automations
- Integrations — Tool connection docs and API notes
Slash commands for Claude Code live in .claude/commands/ at the project root (a Claude Code technical requirement). These are the reusable task definitions Atlas executes when you invoke /new-client, /draft-content, etc.
The CLAUDE.md file at the project root configures Claude Code to operate as Atlas. It points to the Atlas Brief for full personality details and contains the quick-reference rules needed for every session.
For the full Atlas personality and operating principles, see 08 Atlas/Atlas Brief.md.
Client folder setup
When onboarding a new client:
- Create a folder in
01 Clients/using the company name from HubSpot - Create subfolders:
Discovery + Onboarding,Brand,Strategy,Communications,Reporting,Projects,Reference Library, andContent(if applicable). Initiative-specific work goes inProjects/[Project Name]/(each with aProject Indexholding the HubSpot Project Record ID); reusable/reference copies go inReference Library/. - Copy
03 Templates/Client Index — TEMPLATE.mdinto the new client folder - Rename to
Client Index — [Client Name].mdand fill it out - Link the repo client folder path and, if media is in scope, the
WeLaunch Drive/Clients/[Client]/folder URL to the company record in HubSpot - Create a Figma project for the client (if design work is in scope)
- Update the Client Index with the Figma link and all external resource URLs
In Claude Code, use /new-client to automate this process.
Principles
- One home for everything. Don't duplicate information across tools. Point to it.
- The repo is for documents, not data. If it's a contact, a deal, or a metric, it belongs in HubSpot.
- Draft in the repo, publish in HubSpot. Content flows one direction.
- Client folders accumulate, the OS evolves. Specific outputs go to clients. Reusable learnings get extracted to the system.
- Templates stay clean. Copy, rename, then populate. Never edit the original.
- Name things for agents. Consistent, parseable file and folder names make automation possible.
- Standardize client folders. Every client gets the same structure. No snowflakes.
This is a living document. Update it as the OS evolves.