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

# Research what is missing

> Choose how to investigate one open question and verify the evidence you find.

<div className="course-meta" aria-label="Lesson information">
  <span><Icon icon="book-open" /> AI Now · Lesson 4 / 5</span>
  <span><Icon icon="clock-3" /> \~5 min</span>
  <span><Icon icon="user-round" /> Intermediate</span>
</div>

<div className="learning-stages learning-stages-progress" aria-label="Course progress">
  <div className="learning-stage learning-stage-done">
    <span className="learning-stage-mark" aria-hidden="true" />

    <span className="learning-stage-title">Question</span>
  </div>

  <div className="learning-stage learning-stage-done">
    <span className="learning-stage-mark" aria-hidden="true" />

    <span className="learning-stage-title">Source</span>
  </div>

  <div className="learning-stage learning-stage-done">
    <span className="learning-stage-mark" aria-hidden="true" />

    <span className="learning-stage-title">Evidence</span>
  </div>

  <div className="learning-stage learning-stage-active" aria-current="step">
    <span className="learning-stage-mark" aria-hidden="true" />

    <span className="learning-stage-title">Investigate</span>
  </div>

  <div className="learning-stage">
    <span className="learning-stage-mark" aria-hidden="true" />

    <span className="learning-stage-title">Save</span>
  </div>
</div>

The evidence brief in Nowledge Mem names what is missing. Choose a way to investigate one gap: your own team for internal facts, or AI Now Deep Research for public evidence.

## What you'll accomplish

<Check>
  You'll record a checked finding, or a clearly unresolved question, with the next step needed to answer it.
</Check>

## Learn with the simulation

On desktop, the simulation opens automatically when you enter this lesson. Follow the popup hints to send a request, read its reply, and check the result. If you close it, click **Start the simulation** under [Your turn](#your-turn) to reopen it. Each lesson prepares its fictional Atlas materials; no model setup is needed.

<Note>
  The simulation runs locally with preset replies. It does not call AI or search the web, and refreshing resets the practice. On mobile, read the steps first and use a desktop browser for the interactive exercise.
</Note>

## Before practising in the real app

Keep the lesson 3 evidence brief and one open question ready. The time estimate covers setting up and reviewing the request; a research run may take longer.

<Note>
  Use a general question for public research. Keep confidential details and personal data out of web search requests.
</Note>

## Your turn

Follow the hints in the automatically opened simulation. After practising, use the steps below to repeat the workflow with your own material in Nowledge Mem.

<script src="/course-playground-guide.js?v=26" defer />

<div className="course-playground-workflow" data-course-playground-guide="" data-course-guide-autostart="true" data-course-guide-scenario="ai-research" data-course-guide-current="1" data-course-playground-open="false" data-course-guide-replay-label="Practise again" data-course-guide-close-label="Close simulation" data-course-guide-mini-label="Playground" data-course-guide-fill-label="Fill the example" data-course-guide-done-message="Lesson complete. Continue to the next lesson, or try it in Nowledge Mem." data-course-guide-open-mem-label="Open Nowledge Mem" data-course-guide-confirm-label="Next lesson" data-course-guide-confirm-href="/ai-now/save-the-result#simulation" data-course-guide-mem-url="https://mem.nowledge.co" data-course-guide-hint-1="Open AI Now. The sample conversation from lessons 1–3 includes the evidence brief and open questions." data-course-guide-hint-2="Enable Research. This demo does not access the web." data-course-guide-hint-3="Ask the scoped public-research question." data-course-guide-hint-4="Send to see a simulated no-evidence outcome." data-course-guide-hint-5="Lesson complete. Continue to the next lesson, or try it in Nowledge Mem." data-course-guide-fillable-3="Find up to two documented methods that help teams spot cross-team dependencies before committing plans. Prefer original studies, documented trials, or first-hand case reports. For each method, give its source link, publication date, participants, planning horizon, and what was measured. Separate measured outcomes from advice or opinion. Explain what could inform one review of a twelve-month plan and what would still need an internal trial. If you find no relevant evidence, say so. Do not treat our fictional Atlas interviews as public evidence.">
  <button type="button" className="course-playground-launch" data-course-guide-open aria-controls="course-playground-ai-4-en" aria-expanded="false">
    <Icon icon="panel-right-open" /> <span data-course-guide-open-label>Start the simulation</span>
  </button>

  <Steps>
    <Step title="Choose one open question">
      Decide what kind of evidence could answer it. Team capacity needs an internal estimate or trial; external articles cannot establish it. For that route, record who could answer and what to measure, then continue to lesson 5.

      For the public-research route, look for documented methods that help teams find cross-team dependencies before committing plans.
    </Step>

    <Step title="Enable Deep Research for a public question">
      <span className="course-playground-step-anchor" data-course-guide-source="1" aria-hidden="true" />

      <span className="course-playground-step-anchor" data-course-guide-source="2" aria-hidden="true" />

      For this multi-source exercise, turn on **Research** in the AI Now conversation. AI Now can also use web search in ordinary tasks; waiting until now is this course's way of separating saved evidence from new research. If Research is unavailable, record the gap and next investigation step. Mark the research exercise as not run.
    </Step>

    <Step title="Ask for sources, not a generic verdict">
      <span className="course-playground-step-anchor" data-course-guide-source="3" aria-hidden="true" />

      Keep the research scoped:

      ```text theme={null}
      Find up to two documented methods that help teams spot cross-team dependencies before committing plans. Prefer original studies, documented trials, or first-hand case reports.

      For each method, give its source link, publication date, participants, planning horizon, and what was measured. Separate measured outcomes from advice or opinion. Explain what could inform one review of a twelve-month plan and what would still need an internal trial. If you find no relevant evidence, say so. Do not treat our fictional Atlas interviews as public evidence.
      ```
    </Step>

    <Step title="Compare the research with your own evidence">
      <span className="course-playground-step-anchor" data-course-guide-source="4" aria-hidden="true" />

      Open the most relevant cited source. Check the claimed method or result against the original. Compare its participants and planning horizon with the Atlas question. Advice may suggest something to try; it does not prove the review will work for Atlas. If a source cannot be read, leave its claim unverified.

      In the conversation, record the finding, its source and date, the limitation, and what remains unknown. If the run found no relevant evidence, record that limited search outcome and plan a small internal test. Do not turn “not found in this run” into “no evidence exists.”
    </Step>
  </Steps>

  <p className="course-playground-workflow-status" data-course-guide-ready>Follow the popup hints; reopen the simulation if you closed it.</p>
  <p className="course-playground-workflow-status course-playground-workflow-status-complete" data-course-guide-done role="status" hidden>Lesson complete. Continue to the next lesson, or try it in Nowledge Mem.</p>

  <div id="course-playground-ai-4-en" className="course-playground-window" data-course-playground-window role="region" aria-label="AI Now practice" aria-hidden="true">
    <link rel="stylesheet" href="/playground.css?v=3" />

    <div data-mem-playground />

    <script src="/playground-ai-now.js?v=4" defer />

    <script src="/playground.js?v=3" defer />
  </div>
</div>

## What you should see

<Check>
  You have an explicit gap and a next action. If you ran Deep Research, you also have a finding checked against its source, or a record that this run found no relevant evidence. Neither outcome proves that the review works or fails.
</Check>

<Note>
  Planning an internal follow-up completes the investigation-planning route. The Deep Research exercise is complete only after a real run and source review, or a recorded no-evidence result. If sources could not be checked, keep that verification pending.
</Note>

## Why this matters

Research is most useful after you know what is missing. Your own memories and source material set the question; current public sources help fill a specific gap.

## Go deeper

Read the [AI Now Deep Research guide](https://mem.nowledge.co/docs/ai-now) for how AI Now searches and cites multiple sources.