Knowledge Base

The Knowledge Base is Foundaro's institutional memory layer. It gives you a single, structured place to capture and organise everything your company knows — how decisions get made, what processes are followed, what the market context is, and what templates and research you rely on.

The module is built around the idea that knowledge is most useful when it is findable. Documents that live in email threads, Notion pages, or someone's head are not company knowledge — they are individual memory. The Knowledge Base turns individual context into shared, persistent, searchable infrastructure.

Every document is typed by kind, tagged for filtering, linked to related documents, and exportable as Markdown or PDF. The AI assistant lets you query across the entire library in natural language, grounded in your actual documents rather than generic web knowledge.

Overview

The Overview tab is the at-a-glance summary of your knowledge library.

Pinned documents — If you have starred any documents, they appear in a dedicated Pinned section at the top of the overview. This is designed for the 3–5 documents founders reference constantly — your company context, pricing policy, onboarding playbook, and so on. Click any pinned title to open it directly. Click the star icon to unpin.

Recent documents — The six most recently updated documents, with kind and last-updated date. Click any title to preview the full document content. For File kind documents, clicking opens the file directly. For Link kind documents, clicking opens the URL. If you have more than six documents, a link at the bottom takes you straight to the Documents tab.

By kind — A breakdown of how many documents you have in each category. Click any row to jump to the Documents tab filtered by that kind. This is the fastest way to audit whether you have enough policies, playbooks, or research in a given area.

Tips:

• Pin your most-referenced documents so they're always one click away from the overview.
• Use the overview at the start of each quarter to check that your foundational context documents are still current.
• The "By kind" breakdown helps you spot gaps — if you have 15 research documents but no playbooks, your knowledge base is heavy on inputs but light on process.

Documents

The Documents tab is a full-featured document management workspace. You can search, filter, sort, switch layouts, select in bulk, and paginate through your entire library.

Layouts — Three layout options are available, toggled by the icon buttons in the toolbar:

• Cards — A grid of cards showing title, kind badge, status badge, tags, summary, references, and last updated date. Best for browsing at a glance. Shows 12 documents per page.
• List — A compact list showing the same fields plus linked document count and reference count. Best for quick scanning. Shows 20 per page.
• Table — A tabular view with title, kind, status, and updated columns. Best for comparing or auditing a large set. Shows 20 per page.

The selected layout is saved to your browser's local storage and restored on your next visit.

Filtering and sorting:

• Search — Searches across title, summary, content, tags, and URLs using a relevance ranking algorithm. Matching documents are sorted by relevance; the sort dropdown is disabled while a search query is active.
• Kind filter — Narrows the list to documents of a specific kind.
• Pinned toggle — Shows only starred (pinned) documents when active. Turns amber when filtering.
• Sort — When no search query is active, sort by Updated (default), Pinned first, Title A–Z, or Kind.

Actions on every document:

• Preview (click the title) — Opens a full-content preview dialog. For File kind, clicking downloads/opens the file. For Link kind, clicking opens the URL in a new tab.
• Star (star icon) — Pins or unpins the document. Pinned documents appear in the Pinned section on the Overview tab.
• Edit (pencil icon) — Opens the edit dialog to modify all document fields.
• Download (arrow icon) — For standard documents: a dropdown to export as Markdown (.md) or styled PDF. For File kind: opens the uploaded file. For Link kind: opens the URL.
• Delete (trash icon) — Prompts for confirmation before permanently deleting the document and any uploaded file.

Preview dialog — Clicking a document title opens a full preview showing rendered Markdown content (or the URL for Link kind), followed by:

• Connections — Related documents (outbound links, › chevron) and backlinks — documents that link to this one (inbound, ‹ chevron, "links here" label). Clicking any connection navigates to that document's preview.
• References — The full list of external links and internal references attached to the document, with labels, summaries, and clickable URLs.

References on cards — Up to three references from a document are shown directly on its card, so you can open external links without opening the full preview. A "+N more" label appears if there are additional references.

Document fields:

• Title — Required. The document's display name.
• Kind — Required. The document category (see Document Kinds below).
• Status — active, draft, or archived. Only active documents are included in AI search by default.
• Summary — A one or two sentence description. Used in card and list views and in AI context assembly.
• Tags — Comma-separated labels for filtering and discovery. Tags are searchable.
• Related documents — Links to other documents in the knowledge base. Used to build the Knowledge Graph.
• References — External links or internal references with optional labels and summaries. One per line in the format label | url | summary.
• Content — The full document body in Markdown (not shown for File or Link kind).
• URL — Required for Link kind. The external URL this document points to.
• File — Required for File kind. The uploaded file (PDF, image, Word, Excel, PowerPoint, or plain text, max 50 MB).

Bulk edit — Click the Select button in the toolbar to enter selection mode. Checkboxes appear on every card, list row, and table row. A floating action bar appears at the bottom of the screen when at least one document is selected:

• Status — Set all selected documents to Active, Draft, or Archived in one action.
• Pin / Unpin — Star or unstar all selected documents.
• Delete — Permanently delete all selected documents (with confirmation). Uploaded files are cascade-deleted.
• Clear — Deselect without leaving selection mode.

A Select all / Clear bar above the list lets you select the entire filtered set at once. Leaving select mode clears all selections.

Starters — A row of quick-start buttons below the toolbar lets you create a new document pre-filled with a recommended title, kind, summary, tags, and sample content for common document types.

Pagination — Documents are paginated (12 per page for cards, 20 for list and table). The page resets to 1 whenever you change a filter or search query.

Tips:

• Write a summary for every document. It appears in AI search results and helps reviewers decide whether the full document is relevant without opening it.
• Use the Pinned filter to do a quick review of just your most important documents.
• Use the table layout for audits. Sorting by Kind and scanning status badges quickly reveals which documents are still in draft or have been archived.
• Use bulk edit to archive or delete a batch of outdated research notes at the end of a quarter — faster than deleting one by one.
• References on cards make it easy to jump to a Google Doc or Notion page without opening a preview dialog.

Knowledge Graph

The Knowledge Graph shows how your documents connect to each other. It is the structural view of your knowledge base — useful for understanding which documents are foundational, which are isolated, and where the gaps in your internal linking are.

Navigating the graph:

• Select a document from the list on the left to inspect its connections in the detail panel on the right.
• The detail panel shows the document's title and summary, followed by its outbound and inbound connections.

Outbound links (Related) — Documents that the selected document explicitly links to via the "Related documents" field. These are connections you define when editing a document. Click any related document in the panel to navigate to it.

Inbound links (Backlinks) — Documents that reference the selected document. These are computed automatically — you do not need to update a document when something new links to it. A document with many inbound links is a hub document — central to your knowledge structure and worth keeping well-maintained.

References — External links and internal document references attached to the selected document. These are displayed in the detail panel below the connected documents.

Coverage signals:

• A document with zero outbound and zero inbound links is isolated. It exists in the library but cannot be discovered through navigation.
• A document with many inbound links but few outbound links is a hub — a foundational document that others depend on.
• A document with many outbound links but few inbound links is a leaf — a specialised document that builds on a lot of context but isn't yet referenced by others.

Tips:

• Build connections deliberately. When you create a new policy, link it to the context document that explains the operating environment it sits within. When you create a playbook, link it to the policy it implements.
• After adding several documents, spend 15 minutes in the Knowledge Graph reviewing which documents are isolated. Add links to surface them.
• Hub documents — those with many inbound links — should be reviewed more frequently. A stale hub document cascades incorrect context across everything that links to it.
• Use the Edit button in the detail panel to add related document links without leaving the graph view.

Activity

The Activity tab shows all documents in the knowledge base ordered by most recently updated — a simple, paginated log of what has changed and when.

Reading the activity list:

Each row shows the document title, kind label, status badge, and the date it was last updated. The list is sorted by updatedAt descending, so the most recently touched documents appear at the top.

Pagination — 20 documents per page. Navigation controls appear at the bottom when there are more than 20 documents.

Use cases:

• Weekly review — At the start of each week, check which documents were updated in the last seven days. Confirm that the updates were intentional and that nothing important was accidentally changed.
• Staleness audits — Scroll to the bottom of the activity list to find documents that haven't been touched in a long time. Documents that haven't been reviewed in 90+ days should be assessed for accuracy.
• Onboarding — When bringing on a new team member, the activity list helps you identify which documents are actively maintained vs. which are historical.

Tips:

• Archive documents you no longer use rather than leaving them with an old updatedAt date. An archived document is clearly marked as inactive; an outdated active document can mislead.
• If you're doing a knowledge base review, sort by the oldest documents (work backwards from the last page) to find the ones most likely to need updating.
• Set a recurring reminder — monthly or quarterly — to review documents older than 90 days. Knowledge bases go stale quietly, not all at once.

AI Search

The "Ask AI" button in the header opens an AI search dialog. You type a question in natural language, and Foundaro retrieves the most relevant documents from your knowledge base and uses them as grounding context for a Gemini-powered answer.

How it works:

1. You enter a question (e.g. "What is our policy for customer refund requests?").
2. Foundaro searches across all active documents using a relevance ranking algorithm, selecting up to six sources.
3. If no relevant documents are found, the search stops and you are told to add more knowledge first.
4. If sources are found, the selected documents are passed as context to the AI with your question.
5. The AI returns a grounded answer and the answer is displayed in the Overview tab along with the source document titles.

The AI only knows what you've written — it will not draw on general internet knowledge. If you ask about a policy that doesn't exist in your knowledge base yet, the AI will either tell you there are no relevant documents or give a vague answer. The quality of the AI's answers is directly proportional to the quality of your documents.

Source documents — The source titles shown below the AI answer tell you which documents contributed to the response. If an answer seems wrong or incomplete, open the source documents to verify whether the underlying content is accurate.

Clearing the result — Click the ✕ button in the top right of the AI answer card to dismiss it and return the overview to its default state.

Tips:

• Write your documents to answer the questions you'd actually ask. A policy document that starts with "Our refund policy is..." is more useful to AI search than one that starts with a history of why the policy was created.
• Use full sentences in summaries. The AI uses summaries to rank relevance — vague summaries reduce the quality of source selection.
• If an AI answer is confidently wrong, open the source documents it cited and correct the content. The AI cannot fix its own sources.
• AI search uses 2 AI credits per query. Plan queries for questions where you genuinely don't know which document holds the answer — not for documents you could find by searching manually.

Document Kinds

Every document in the Knowledge Base is assigned a kind. Kind is the primary structural category — it controls filtering, display, and how the AI ranks documents for relevance in certain query types.

Context — Background documents that explain the operating environment, market, history, or company positioning. These are evergreen documents that don't change frequently but inform everything else. Examples: company overview, market positioning, ICP summary, origin story, operating principles.

Policy — Rules and guidelines that govern decisions. Policies are authoritative — they define what is and isn't allowed. Examples: refund policy, hiring policy, pricing authority, data retention policy, code of conduct.

Playbook — Step-by-step processes for recurring activities. Playbooks describe how things are done, not just what the rules are. Examples: customer onboarding playbook, incident response playbook, sales outreach sequence, quarterly review process.

Template — Reusable structures for documents, emails, proposals, meeting agendas, or other recurring output formats. Templates are meant to be copied and filled in. Examples: board update template, investor update template, customer proposal template, job description template.

Research — External findings, market analysis, competitor notes, and customer interview summaries. Research documents are often time-sensitive — they capture what you knew at a specific point in time. Examples: competitor analysis (Q1 2026), user interview synthesis, market sizing model notes, industry report summary.

File — An uploaded file attached to the knowledge base. Supports PDF, images, Word, Excel, PowerPoint, and plain text up to 50 MB. The title click and action button open the file directly — there is no content editor. Use File kind for documents that exist in their original format and don't need to be rewritten in Markdown. Examples: signed contracts, slide decks, compliance certificates, design exports.

Link — An external URL registered as a knowledge document. Title click and action button open the URL in a new tab. The URL is shown on the card and in the preview. Use Link kind to register important external resources — Google Docs, Notion pages, Loom recordings, Figma files, Airtable bases — so they are discoverable from the knowledge base without copying content. Examples: live product roadmap in Notion, engineering runbook in Confluence, onboarding Loom video, brand guidelines in Figma.

Choosing the right kind — If a document both defines a rule and explains how to follow it, use Policy for the rule and Playbook for the process. If a document contains both research findings and a strategic recommendation, split them — keep the raw findings as Research and the recommendation as a Context or Policy document. Use File for source-of-truth assets that shouldn't be paraphrased. Use Link for living external documents that are maintained elsewhere.

Export

Every standard document can be exported from the download icon in the document actions. File and Link kind documents open their content directly rather than offering an export menu.

Markdown export — Downloads the raw document content as a .md file. The filename is derived from the document title. Use Markdown export when:

• You want to import the content into another tool (Notion, Linear, GitHub Wiki).
• You want to share the raw text with someone outside Foundaro.
• You want to version the document in a code repository.

PDF export — Generates a styled PDF with the document title and kind/date subtitle in the header, followed by the rendered Markdown content. The PDF uses a serif font for readability and handles page breaks around headings, lists, and code blocks. Use PDF export when:

• You are sharing a polished document externally (e.g. a proposal or policy document for a partner).
• You need a printable, fixed-layout version for a meeting or board packet.

Size limit — Individual document content is limited to approximately 900 KB. Very large documents will be rejected at save time with an error message. Split large documents into multiple linked sections and use the Knowledge Graph to connect them.