Knowledge Base
The Knowledge Base is your asset's memory: The more it knows, the more specific and on-brand every recommendation and draft becomes.
What to add
Seed voice, positioning, and product facts first. Mission Brief and Content Operations retrieve from the Knowledge Base on every run, so density here sharpens every downstream output.
- Brand voice and positioning.
- Product or feature catalog.
- Ideal customer profile and personas.
- Competitor facts you want outputs grounded in.
How it's used
Documents are indexed into embeddings and retrieved when the AI Analyst answers a question or Content Operations writes a draft. This grounding is what keeps outputs accurate to your business instead of generic.
Your Knowledge Base grounds your own outputs. It is not used to train public models.
Import sources
- Paste text: fastest during onboarding.
- Bulk / Obsidian vault: drag-drop or folder-select .md files (5 credits per new file, max 20 per run).
- GitHub: sync markdown from a connected repo (Obsidian Git plugin).
- Google Drive: connect in Integrations, pick a folder, run explicit import.
- Slack: connect in Integrations, select channels, import pinned strategy threads.
- Feed voice docs first so Content Operations drafts sound on-brand on day one.
Imports are explicit only. Nothing syncs silently on connect. Re-imports skip unchanged paths via metadata dedupe.
Obsidian setup
- 1Local vault: Knowledge → Bulk import → select your vault folder (.md files). Unchanged files are skipped on re-import; edits re-embed automatically.
- 2CLI one-shot: learn-domains knowledge bulk --website yourdomain.com --dir ./vault
- 3Portfolio CLI: learn-domains knowledge sync init → edit ~/.config/learn-domains/obsidian-sync.json → learn-domains knowledge sync run
- 4Git-backed vault: push with Obsidian Git → connect GitHub in Integrations → Import from GitHub (sha-based incremental sync).
- 5Seed voice, ICP, and pricing docs first. Dense beats dumping every daily note.