TL;DR: The template is ready. This walkthrough shows you how to fill every section using a real company example. By the end, you’ll know exactly what goes where and why.
So you’ve finally convinced everyone. Now the fun stuff begins. Clone the migration repository from GitHub into what’ll be the new folder where everything will be. Use git clone https://github.com/markdown-company/ai-native-migration-templates.git or download as ZIP from the repo page.
For the migration you need a plan. An airtight plan. Something that’ll mention every decision taken, every change, everything. Because you need a base to get it approved by your colleagues, of course, but more pragmatically because it’ll be an AI doing the job. So work on the plan with AI. Ask the AI to review the plan many times. Let the AI execute.
What is the migration plan template?
It’s a fill-in-the-blanks document at migration-toolkit/plan-templates/Migration Plan Template.md in the migration toolkit. Copy it into your own 00-Migration Control/ folder, then work through each section with your team.
The template covers:
- Why you’re migrating (purpose, business context)
- What you’re building (target structure, access model, folder tree)
- How you’ll get there (mapping, conversion, step-by-step process)
- How you’ll know it’s done (validation checklist)
The rest of this walkthrough uses a mock company — Stellar Robotics — so you can see what a filled-in plan looks like before you write your own.
Purpose & business context
These two sections set the stage. Keep them short — one sentence each is enough.
Purpose — one line that says what you’re doing and why.
Define how to reorganize Stellar Robotics’ files into an AI-ready, access-aware folder system. Goal: make company knowledge easier to search, link, and use with AI tools — while keeping access boundaries clear.
Business Context — capture the date, your main product, and any legacy products the AI needs to know about.
| Field | Example |
|---|---|
| Current Date | 2026-06 |
| Primary Product | Stellar Navigation Platform |
| Legacy Products | Stellar Tracker v1, Stellar Analytics |
The AI uses this context to understand that file names like “Tracker v1 budget” belong to a legacy product, not current operations.
Core decisions
This is the most important section. These 17 decisions define how the migration works. Present them to your team, get agreement, then move on.
The most impactful ones:
| # | Decision | Why it matters |
|---|---|---|
| 1 | Migrate into a new root folder first | You never touch the old tree. If something goes wrong, nothing is lost. |
| 2 | Files are copied, not moved | Same reason — safety. Delete the old tree only after validation. |
| 3 | Keep existing sync (Nextcloud, SharePoint) | Don’t break what works. Git handles the markdown; the old sync handles the rest. |
| 4 | Access-aware vaults, not one giant vault | The AI can’t accidentally see HR files if they’re in a different folder. |
| 5 | No frontmatter on every note by default | Frontmatter is powerful but adds friction. Use it only where filtering matters. |
| 6 | Backlinks > metadata for most notes | A [[link]] is faster to write and easier to maintain than a tag system. |
The full list is in the template. Read through each one with your team and mark your agreement.
Root structure & access model
These two sections are the skeleton of your vault. Define them early — everything else hangs off them.
Stellar Robotics’ root structure:
Stellar Robotics Reorganization/
+-- 00-Migration Control/
+-- 01-Main Vault/
+-- 02-Legal Vault/
+-- 03-HR and GDPR Vault/Access model — who sees what:
| Vault | Who can access | What goes in |
|---|---|---|
| 01-Main Vault | Everyone, day-to-day | Products, relationships, investors, marketing, research, ops |
| 02-Legal Vault | Counsel + leadership | Contracts, signed agreements, NDAs, IP filings |
| 03-HR and GDPR | HR only | Employment docs, CVs, team records, GDPR |
After validation and approval, rename “Stellar Robotics Reorganization” to the final vault name (e.g., “Stellar Robotics Vault”).
Folder mapping
This is the execution tool. Before copying a single file, map every source folder to its destination.
Create a table in 00-Migration Control/Folder Mapping.md:
| Source | Destination | Action | Notes |
|---|---|---|---|
SharePoint/Marketing/Brand | 01-Main Vault/04-Marketing/Company/Brand/ | copy | Brand guidelines |
Google Drive/Sales/Deals | 01-Main Vault/02-Relationships/Pipelines/ | convert | Convert spreadsheet to markdown table |
CEO Office/Board Updates | 01-Main Vault/03-Investors/Pitch Materials/ | copy | |
HR/Payroll | 03-HR and GDPR Vault/Employment Documents/ | copy | Confidential. |
Use four action types:
- copy — straight copy, no conversion
- convert — convert binary to markdown using the toolkit’s
./convertcommand - reference — leave in place, create a link/note pointing to it
- manual review — unclear destination, needs human decision
The mapping is your single source of truth during the copy phase. Review it before you start. Update it as you find edge cases.
Folder responsibilities
Each top-level folder gets a one-line job description. This prevents scope creep — someone can’t argue “put the budget in Marketing” because Marketing’s responsibility is already defined.
| Folder | Responsibility |
|---|---|
| 01-Products | Product-specific knowledge: notes, app, launch, feedback, testing, budgets |
| 02-Relationships | CRM-ready company and individual records, pipelines, interactions |
| 03-Investors | Investor CRM, pitch materials, data room, financial planning |
| 04-Marketing | Public-facing assets: brand, website, social, testimonials, flyers |
| 05-Research | Scientific papers, technical research |
| 06-Market Research | Competition, market info, business use cases, industry notes |
| 07-Operations | SOPs, templates, meeting notes, process docs, handovers |
| 02-Legal Vault | Contracts, signed agreements, NDAs, IP, corporate docs |
| 03-HR and GDPR | Employment, CVs, hiring, team records, GDPR documents |
Key details (grouped)
These template sections are important but don’t need a full walkthrough. Here’s what you need to know about each:
Exclusions and Manual Review — Some folders don’t belong in the new vault. If a folder contains confidential documents stored elsewhere (e.g., a lawyer’s secure drive), mark it exclude or reference. Track every exclusion in the mapping. If you’re unsure, mark it manual review and decide later.
.gitignore — Add a .gitignore in the migration repo root to keep searches focused:
_Archive - DO NOT EDIT/
OLD*/
ARCHIVE*/
node_modules/
.obsidian/
.trash/
.DS_StoreWithout it, every search scans thousands of binary files. An AI searching 5,000 source files plus 20,000 archived documents returns noise. An ignore pattern returns signal.
Naming Convention — Date-first where chronology matters: YYYY-MM-DD - Subject. Don’t rename files unnecessarily — only where the timeline adds meaning (e.g., funding applications, project milestones).
Duplicate Handling — Run exact-duplicate detection (by file hash) before and after migration. Remove only exact duplicates. Never auto-delete near-duplicates, drafts, or different versions. Keep the report in 00-Migration Control/Duplicate Reports/.
Conversion Tools — The toolkit includes a ./convert CLI that handles .docx, .doc, .xls, .xlsx, and .pptx. Run it during the copy phase:
./convert document.docx -o output.md
./convert spreadsheet.xlsx --summary -o output.mdOCR Policy — OCR (converting scanned images to searchable text) is a later-stage enrichment. You don’t have to do it in the first migration pass. But if you want to do it already, see src/ocr_batch.py
Sync and Versioning — Keep your existing sync system (Nextcloud, SharePoint) during migration. If you switch to Git, use it once the migration is concluded.
Relationship & investor models (optional)
The template includes lightweight markdown CRM templates for companies and individuals. One .md note per important entity, connected with backlinks for the CRM, and with frontmatter for Investors. Choose the model that suits better for each case.
For Stellar Robotics, this means:
- Company note for each partner, customer, and vendor
- Individual note for each contact person
- Backlinks connecting companies → people → meetings → products
The same structure works for investors, with a separate template that includes frontmatter for pipeline stage, last contact, and next action.
These are optional — add them only where relationship tracking adds value. Relationships and Investors are the most common use cases.
Validation checklist
The template ends with a comprehensive checklist (~30 items). Walk through it in a team meeting after the copy phase.
Key checkpoints:
- Files copied, not moved
- Product material copied (converted from binary where appropriate)
- Legal/signed documents in Legal Vault
- HR/private data in HR and GDPR Vault
- Exact duplicate report generated
- Completeness validation run (original vs. reorganized file count)
- Access boundaries reviewed
- Final approval received
- Old folder tree deleted after approval
- Reorganization root renamed after approval
The full checklist is in the template. Print it, check it, sign it.
Open questions
The template ends with 7 example questions. Let the AI find its own questions during reviews.
Involve your team
Spend time with your team. You’ll get a finer reorganization, but more critically you’ll figure out where the resistance is the strongest. Focus your efforts there when training.
What’s next
Plan is ready. Now convert your files: Convert DOCX, XLS, and PPTX to Markdown — turn Office files into plain text.
Markdown your next move. By Charles Henri Gayot.