> ## Documentation Index
> Fetch the complete documentation index at: https://academy.pathfindr.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Building Your Knowledge Base

> Give AI persistent context about your business so it stops being generic and starts being genuinely useful.

export const Quiz = ({question, options, correctIndex, explanation}) => {
  const [selected, setSelected] = useState(null);
  const [showResult, setShowResult] = useState(false);
  const handleSelect = index => {
    if (showResult) return;
    setSelected(index);
    setShowResult(true);
  };
  const isCorrect = selected === correctIndex;
  return <div className="my-8 p-6 rounded-2xl border border-zinc-200 dark:border-zinc-700 bg-gradient-to-br from-zinc-50 to-white dark:from-zinc-900 dark:to-zinc-800 shadow-sm">
      <div className="flex items-center gap-3 mb-4">
        <div className="flex items-center justify-center w-10 h-10 rounded-xl bg-blue-100 dark:bg-blue-900/40">
          <span className="text-xl">📝</span>
        </div>
        <div className="flex flex-col">
          <p className="font-semibold text-zinc-800 dark:text-zinc-100 m-0">Knowledge Check</p>
          <p className="text-xs text-zinc-500 dark:text-zinc-400 mt-0.5 m-0">Select the best answer</p>
      </div>
      </div>
      
      <p className="text-zinc-700 dark:text-zinc-200 font-medium mb-5 text-lg leading-relaxed">{question}</p>
      
      <div className="space-y-3">
        {options.map((option, index) => {
    const isSelected = selected === index;
    const isCorrectOption = index === correctIndex;
    let optionStyles = "border-zinc-200 dark:border-zinc-600 hover:border-blue-400 hover:bg-blue-50 dark:hover:bg-blue-900/20";
    if (showResult) {
      if (isCorrectOption) {
        optionStyles = "border-green-500 bg-green-50 dark:bg-green-900/30";
      } else if (isSelected && !isCorrectOption) {
        optionStyles = "border-red-500 bg-red-50 dark:bg-red-900/30";
      } else {
        optionStyles = "border-zinc-200 dark:border-zinc-700 opacity-60";
      }
    }
    return <button key={index} onClick={() => handleSelect(index)} disabled={showResult} className={`w-full text-left p-4 rounded-xl border-2 transition-all duration-200 ${optionStyles} ${!showResult && 'cursor-pointer'} ${showResult && 'cursor-default'}`}>
              <div className="flex items-start gap-3">
                <div className={`flex-shrink-0 w-6 h-6 rounded-full border-2 flex items-center justify-center text-xs font-bold ${showResult && isCorrectOption ? 'border-green-500 bg-green-500 text-white' : showResult && isSelected && !isCorrectOption ? 'border-red-500 bg-red-500 text-white' : 'border-zinc-300 dark:border-zinc-500 text-zinc-500 dark:text-zinc-400'}`}>
                  {showResult && isCorrectOption ? '✓' : showResult && isSelected ? '✗' : String.fromCharCode(65 + index)}
                </div>
                <span className="text-sm text-zinc-700 dark:text-zinc-200 leading-relaxed">{option}</span>
              </div>
            </button>;
  })}
      </div>
      
      {showResult && <div className={`mt-5 p-4 rounded-xl ${isCorrect ? 'bg-green-100 dark:bg-green-900/40 border border-green-200 dark:border-green-800' : 'bg-amber-100 dark:bg-amber-900/40 border border-amber-200 dark:border-amber-800'}`}>
          <div className="flex items-start gap-3">
            <span className="text-xl">{isCorrect ? '🎉' : '💡'}</span>
            <div className="flex flex-col">
              <p className={`text-sm font-semibold m-0 mb-1 ${isCorrect ? 'text-green-800 dark:text-green-200' : 'text-amber-800 dark:text-amber-200'}`}>
                {isCorrect ? 'Correct!' : 'Not quite right'}
              </p>
              <p className="text-sm text-zinc-600 dark:text-zinc-300 m-0 leading-relaxed">{explanation}</p>
          </div>
          </div>
        </div>}
    </div>;
};

<img src="https://mintcdn.com/pathfindr/I3XbhzOlK6BeuXVO/images/LCWeek2Notebook.webp?fit=max&auto=format&n=I3XbhzOlK6BeuXVO&q=85&s=12d69eee73eb82ab7202dc1568283610" alt="leadership cohort notebook Pn" noZoom={true} width="1920" height="1080" data-path="images/LCWeek2Notebook.webp" />

## What you will learn

Every time you start a new AI chat, it knows nothing about you. This section shows you how to fix that — building a knowledge base that gives AI the context it needs to produce relevant, specific output from the first message.

<CardGroup cols={4}>
  <Card title="Why context matters" icon="square-1">
    Understand why AI outputs are generic by default and what changes when you give it your world
  </Card>

  <Card title="NotebookLM" icon="square-2">
    Use Google's NotebookLM to create a queryable knowledge base from your own documents
  </Card>

  <Card title="Memory and instructions" icon="square-3">
    Use built-in memory and custom instructions to carry context across every conversation
  </Card>

  <Card title="What to upload" icon="square-4">
    Know which documents belong in a knowledge base and which governance considerations apply
  </Card>
</CardGroup>

***

## Why this matters to you

You have used AI in this session with a scenario and context pasted directly into the prompt. That works. But it requires you to re-explain your business, your priorities, and your context every single time you start a new conversation.

A knowledge base removes that friction. Instead of briefing AI from scratch, you upload your documents once. AI draws on them every time you ask a question. The output stops being generic advice and starts being analysis grounded in your actual business.

<Tip title="The onboarding metaphor">
  Think of building a knowledge base as onboarding a new team member. You would not expect a new hire to give useful input on day one without any background. You would give them the strategy doc, the key client briefs, the recent financials, and the context they need to contribute. A knowledge base does the same thing for AI — permanently.
</Tip>

***

## What a knowledge base adds

AI tools like ChatGPT already carry some context about you if you have set up custom instructions or memory. A knowledge base takes that further. Instead of remembering your preferences, AI can now reference your actual documents, your strategy, and your organisation's specific context in every response.

<CardGroup cols={2}>
  <Card title="Without a knowledge base" icon="xmark">
    Responses draw on general knowledge. Analysis applies to your sector, not your organisation. You find yourself re-explaining background that does not change, your client base, your competitive position, your priorities, each time you start a new task.
  </Card>

  <Card title="With a knowledge base" icon="check">
    Responses reference your actual documents. Analysis is grounded in your strategy and context. The starting point is already calibrated to your organisation. Less time briefing, more time using the output.
  </Card>
</CardGroup>

<div className="mt-8" />

<Accordion title="Test your understanding">
  <Quiz
    question="You ask NotebookLM a question about your competitive position and it references an analysis from six months ago that is now outdated. What is the right response?"
    options={[
  "Trust the output because NotebookLM only draws from your documents, not the internet",
  "Remove the outdated document and upload the current version so the knowledge base reflects your actual position",
  "Switch to standard ChatGPT because it has more up-to-date knowledge",
  "Add a note to the prompt each time reminding AI that the document is outdated"
]}
    correctIndex={1}
    explanation="A knowledge base is only as current as the documents in it. Outdated documents produce outdated analysis. Maintaining your knowledge base — removing old versions and adding updated ones — is part of making it genuinely useful."
  />
</Accordion>

***

## NotebookLM: a knowledge base you can query

Google NotebookLM is purpose-built for this. You upload documents — strategy papers, client briefs, research reports, board presentations, meeting notes — and it creates a knowledge base you can query in natural language.

The key difference from standard AI chat is that NotebookLM only draws on what you have uploaded. It does not blend your documents with general internet knowledge.

<Tip title="One exception worth knowing">
  NotebookLM does have a Discover feature that can search the internet, but this is not part of the standard setup. In most cases, and for everything covered in this session, it works purely from what you have uploaded.
</Tip>

<Steps>
  <Step title="Create a new notebook">
    Open NotebookLM and create a notebook for the project or context you are working on. You can have multiple notebooks — one for a specific deal, one for your business strategy, one for a client account.
  </Step>

  <Step title="Upload your source documents">
    Add the documents relevant to this notebook. Strategy docs, research outputs, client briefs, competitor reports, meeting notes. NotebookLM accepts PDFs, Google Docs, and pasted text. The more relevant the sources, the more useful the outputs.
  </Step>

  <Step title="Ask questions in natural language">
    Query the notebook as you would ask a well-briefed colleague. "What are the three biggest risks across these documents?" or "What does our strategy say about entering new markets?" or "Find any contradictions between the briefing and the research report."
  </Step>

  <Step title="Use the Audio Overview">
    NotebookLM can generate an Audio Overview — a conversational summary of your documents in podcast format. Use this to get across a large volume of material quickly, or to share context with a colleague who does not have time to read the source documents.
  </Step>

  <Step title="Build on the output">
    Use the knowledge base output as an input for your Thinker, Assistant, or Creator prompts. The knowledge base surfaces what is in your documents. The three roles turn it into something you can act on.
  </Step>
</Steps>

***

## Carrying context across conversations

NotebookLM is a dedicated tool for document-based knowledge bases. But most AI tools have built-in ways to carry context across conversations, so AI knows something about you before you say a word. The three most common are custom instructions, memory, and dedicated project spaces.

<AccordionGroup>
  <Accordion title="Custom instructions">
    Most AI tools let you set standing instructions that apply to every conversation. Use these to tell AI who you are, what you do, how you want responses formatted, and what context it should always keep in mind.

    ```text theme={null}
    Example: "I am the CEO of a professional services firm with 200 staff. I work fast and need direct, concise responses. Never use bullet points unless I ask. Always flag risks explicitly."
    ```

    These instructions persist across every new chat. You do not need to re-explain your context each time. Available in ChatGPT, Claude, Copilot, and Gemini.
  </Accordion>

  <Accordion title="Memory">
    Some AI tools store things they learn about you across conversations. You can also manually tell them to remember something: "Remember that our key client is in the aged care sector and we are currently in a competitive pitch process."

    Review your saved memories periodically and remove anything outdated. Memory is useful but not infallible — verify that what has been stored is still accurate. Available in ChatGPT and Claude.
  </Accordion>

  <Accordion title="Project spaces">
    Several AI tools let you create a persistent workspace where you upload documents and set custom instructions. Every conversation within that space starts with that context already loaded. Use this for ongoing work where you want AI to always have background on the topic. Available in ChatGPT (Projects), Claude (Projects), and Copilot (Notebooks).
  </Accordion>
</AccordionGroup>

***

## What to put in your knowledge base

Not every document belongs in a knowledge base. The most useful inputs are documents that would help a well-briefed analyst answer questions about your business, your clients, or your strategic priorities.

<CardGroup cols={2}>
  <Card title="High-value inputs" icon="circle-check">
    Strategy and planning documents, client briefs and account summaries, research reports and market analysis, board presentations and investment theses, project briefs and scopes of work, recent meeting notes with decisions and actions
  </Card>

  <Card title="Lower-value inputs" icon="circle-minus">
    Documents with highly sensitive personal data, outdated materials that no longer reflect current strategy, one-off communications that do not contain reusable context, anything containing information that should not be shared beyond its original recipients
  </Card>
</CardGroup>

<div className="mt-8" />

<Accordion title="Test your understanding">
  <Quiz
    question="A colleague wants to upload a confidential client contract to a personal ChatGPT Plus account to help draft a proposal. What should you tell them?"
    options={[
  "That is fine — ChatGPT Plus accounts are secure",
  "That is fine as long as the client has not explicitly asked for confidentiality",
  "Check your organisation's AI policy and the data handling terms of your specific plan before uploading confidential client documents",
  "Never use AI for any task involving client documents"
]}
    correctIndex={2}
    explanation="Data governance depends on the specific plan, the organisation's policies, and the nature of the document. Personal AI accounts may not offer the same protections as enterprise accounts. The right step is to verify before uploading — not to assume it is safe, and not to avoid AI entirely."
  />
</Accordion>

***

## Try this now

Think about the work you're doing this week. What's one project or ongoing area where AI could deliver much better results if it actually understood your context?

Pick three to five documents that would give AI the background it needs. Upload them to a NotebookLM notebook or paste the key context into your AI tool's memory or custom instructions. Then ask it a question you would normally ask a well-briefed colleague.

Notice what changes.

<Card title="Workflow Discovery Tool" icon="arrow-up-right-from-square" href="/leadership-cohort/week-2-meet-your-exec-ai-assistant/challenge">
  Use this tool to map which parts of your ongoing workflows would benefit most from persistent AI context — and identify the first documents worth uploading.
</Card>

<Tip title="Where this fits on your AI journey">
  A knowledge base is the foundation of every AI agent you will eventually build. Agents do not operate in a vacuum — they draw on context about your business, your clients, and your processes to make decisions and produce relevant outputs. Building your knowledge base now means you are not starting from scratch when you get to agent-building. You are giving your future agents the same context you are giving AI today.
</Tip>

***

## Quick checkpoint

You are done with this section when you can:

<CardGroup cols={4}>
  <Card title="Create a notebook" icon="circle-check">
    Set up a NotebookLM notebook and upload at least three relevant documents
  </Card>

  <Card title="Query your knowledge base" icon="circle-check">
    Ask a question you would normally ask a colleague and compare the output to a standard AI chat response
  </Card>

  <Card title="Set persistent context" icon="circle-check">
    Configure custom instructions or memory in your primary AI tool so it knows your role and context
  </Card>

  <Card title="Apply governance" icon="circle-check">
    Know which documents are appropriate to upload based on your organisation's AI policy and your plan's data handling terms
  </Card>
</CardGroup>