Automating AI trends with Grok

Use Grok live X search to build a sourced, actionable daily AI briefing.

Introduction

30 Second Summary

AI updates race through X faster than most newsletters can catch them. Useful announcements are buried beneath noisy posts.

In this project, you will turn Grok's live X search into a short daily briefing for the AI topics you care about. Your reusable prompt will filter hype through source checks before offering practical next steps.

What You'll Build

You'll run one saved prompt to see a concise briefing that surfaces the recent AI stories worth your attention.

By the end of this project, you'll have:

Build a focused watchlist around two or three AI interests connected to your work or study.
Run a repeatable scan across the past 24 hours of X activity. The result stays short enough to read each day.
Check every story through a status label plus an original-source link. A relevance note shows why it matters to you. A practical action gives you a next move.
Secret Mission: Schedule your finished prompt as a daily Grok Automation when your account supports the feature.

Are there any prerequisites?

You need a Grok account with access to the Grok web app. No AI or technical experience is required.

Before We Start

A useful trend finder starts with a clear purpose. This step helps you decide which updates matter in your day before you begin searching.

Prepare Your Trend Watchlist

Your purpose statement gives your trend finder a destination. A small watchlist keeps the daily scan tied to what you actually do.

In this step, you will confirm access to Grok. You will also choose two or three focused AI interests that match your purpose.

In this step, get ready to:

Access the Grok web app with the chat input ready.
Choose two or three focused interests that support your personal goal.
Record each topic exactly as you want it to appear in later prompts.

Open Grok and confirm access

The Grok web app is where you will run your trend finder. An empty chat input confirms that your account is ready before you build the watchlist.

Use your browser's address bar to visit grok.com.
Sign in with the Grok account you already have.
Confirm that you can see the chat input.
Leave the chat input empty.

That first access check is complete. Grok is ready for the focused topics you choose next.

Can't reach the chat input?

Check that your browser is visiting grok.com.
Reload the page.
Sign in again with your existing account.

Get help with your access issue by selecting Help me troubleshoot access to the Grok web app..

Choose your focused interests

A focused interest describes a specific use of AI for a specific person or task. Examples include AI for Python beginners or AI for graphic design.

Re-read the personal purpose statement you completed earlier.
Choose a first topic connected to something you regularly do.
Choose a second topic covering another important part of your goal.
Add a third topic only if it deserves a regular place in your briefing.

Keep Your Watchlist Focused

Topics tied to real work produce a more useful brief. AI for Python beginners gives Grok a clearer target than AI and technology.

Write your final watchlist

Your topic wording becomes part of the prompts you use later. Recording each topic now keeps that wording consistent whenever you run the trend finder.

Record your first topic here: your first focused AI topic.
Record your second topic here: your second focused AI topic.
Record your optional third topic here if you chose one: your optional third focused AI topic.

Before your final check, what do you expect to confirm about your Grok access and your watchlist?

Switch back to the Grok web app from earlier.
Confirm that the chat input is still visible.
Read your two or three topics aloud exactly as written.
Check that every topic supports your personal purpose.

You should see the empty chat input in Grok. You should also have two or three focused topics written exactly as you want them to appear in your prompts.

Your personal watchlist is ready. Next up, you will use those topics to run a quick live news scan.

Run a Shallow Live-X Scan

Your watchlist now protects the AI topics that matter to your goal. The next challenge is seeing what those topics look like in today's fast-moving feed.

Grok can use live X search to surface recent updates. You will deliberately request a headline-only scan so you can test how useful freshness is on its own.

In this step, get ready to:

Customize the shallow scan with your watchlist topics.
Run the scan across X posts from the past 24 hours.
Inspect the five-headline result for unanswered questions.

Customize your headline-only prompt

A reusable prompt needs the exact topic names from your watchlist. Choose the template that matches the number of topics you selected.

Two Topics

Customize Prompt 1 by filling the two topic fields in the template below:

Use live X search to find AI news from the past 24 hours about [[TOPIC_1="your first watchlist topic"]] and [[TOPIC_2="your second watchlist topic"]]. Return only five short headlines. Do not explain why they matter or suggest any actions.

What does this prompt do?

The topic fields focus the scan on the two interests you selected.
The past 24 hours defines the time window for the search.
The five-headline limit keeps the result short.
The final sentence holds the output to headlines.

Read the rendered prompt after filling both topic fields.
Confirm both topics match your written watchlist exactly.

Three Topics

Customize Prompt 1 by filling the three topic fields in the template below:

Use live X search to find AI news from the past 24 hours about [[TOPIC_1="your first watchlist topic"]], [[TOPIC_2="your second watchlist topic"]], and [[TOPIC_3="your third watchlist topic"]]. Return only five short headlines. Do not explain why they matter or suggest any actions.

What does this prompt do?

The topic fields focus the scan on the three interests you selected.
The past 24 hours defines the time window for the search.
The five-headline limit keeps the result short.
The final sentence holds the output to headlines.

Read the rendered prompt after filling all three topic fields.
Confirm all three topics match your written watchlist exactly.

Topics not appearing correctly?

Check that every topic field contains the exact wording from your watchlist.
Use the two-topic template if your watchlist contains only two interests.
Remove any topic wording that you did not choose.

Help me correct my customized Prompt 1.

✔️ Awesome, I've got everything!

Your customized Prompt 1 is ready to test in a fresh Grok conversation.

ⓧ I'd like to double check the full code

Choose the version that matches your watchlist. Compare every sentence with your rendered prompt.

Two Topics

Compare your two-topic prompt with this complete version:

Use live X search to find AI news from the past 24 hours about [[TOPIC_1="your first watchlist topic"]] and [[TOPIC_2="your second watchlist topic"]]. Return only five short headlines. Do not explain why they matter or suggest any actions.

Three Topics

Compare your three-topic prompt with this complete version:

Use live X search to find AI news from the past 24 hours about [[TOPIC_1="your first watchlist topic"]], [[TOPIC_2="your second watchlist topic"]], and [[TOPIC_3="your third watchlist topic"]]. Return only five short headlines. Do not explain why they matter or suggest any actions.

Run the shallow scan

A prompt becomes useful once you compare its output with your goal. This narrow request gives you a clear baseline to inspect.

Before you send it, do you expect five headlines to feel like enough information for a decision? Keep your prediction in mind.

Start a new Grok conversation using the new-conversation control beside the conversation list.
Paste your customized Prompt 1 into the chat input.
Send the prompt.

You should see five short headlines about your watchlist topics from the requested time window. You now have your first live scan.

Read all five headlines.
Check whether each item shows who originally announced the story.
Check whether each item explains its relevance to your personal goal.
Check whether each item gives you a practical next move.

Why This Result Falls Short

The scan proves that your watchlist can surface recent topic matches. Its headline-only format provides no requested method for checking the evidence behind each claim.

Personal relevance is also absent. Practical next steps are absent too.

That gap is the designed result of this first scan. You now have a baseline that exposes what a stronger briefing must add.

Seeing a Different Format?

Resend the customized prompt without adding extra instructions if Grok includes explanations or actions.
Confirm the prompt contains the instruction to return only five short headlines if the result has a different number.
Check that the time window still says the past 24 hours.

Help me get the intended five-headline result.

Verify the intended limitation

Your scan has done its job when its boundaries are easy to see. The final check confirms that you produced the exact baseline needed for the next refinement.

Before you check, do you expect this result to tell you which headline is trustworthy? Keep your answer in mind.

Count the headlines in the response.
Confirm each headline relates to at least one watchlist topic.
Confirm the scan responds to the past-24-hours request.
Confirm the response contains no source analysis.
Confirm the response contains no relevance explanations.
Confirm the response contains no recommended actions.

You should count five short headlines about your watchlist topics from the requested time window. You should see no source analysis.

Relevance explanations should be absent. Recommended actions should also be absent.

That is the intended win. You have proved that Grok can produce a current scan while preserving a clear baseline for improvement.

Your live scan gives you a short baseline of current headlines. Next, you will trace each story to its source so the evidence becomes visible.

Separate Real News from Hype

Your first scan proved that Grok can surface current X headlines for your watchlist. The result still leaves credibility hidden.

This step makes evidence visible before you act on a headline. You will classify each story by tracing it back to the account responsible for the news.

In this step, get ready to:

Classify the five headlines using evidence from original accounts.
Require a named source plus an original X-post link or a clear disclosure that no primary source was found.
Compare at least one original X post with Grok's summary.

Classify the five headlines

A primary source comes from an account directly responsible for the news. That may be a company account or a product team account.

A project account or researcher account can also qualify. Prompt 2 uses that evidence standard to expose uncertainty in the shallow scan.

Switch back to the shallow-scan conversation from the previous step.
Select the chat input beneath Grok's five headlines.
Paste Prompt 2 into the chat input:

Review the five items you just found.

For each item:
- Label it "Confirmed launch or announcement" only when the original company, product team, project, or researcher account posted the news.
- Otherwise label it "Rumour or hype."
- Name the source account and include a link to the original X post.
- If you cannot find an original source, write "No primary source found."
- Remove anything that falls outside the past 24 hours.

Keep the result short.

How does Prompt 2 improve the scan?

The label Confirmed launch or announcement is reserved for news posted by an account directly responsible for it.
The label Rumour or hype exposes items without that level of evidence.
The phrase No primary source found makes missing evidence visible.
The time filter removes items that fall outside the requested window.

Submit Prompt 2.
Wait for Grok to finish checking the five headlines.

Grok should return a short classified briefing. The raw scan now has visible credibility labels plus source details.

Still seeing a headline-only list?

Confirm that you submitted Prompt 2 in the same conversation as the five headlines.
Compare your submitted prompt with the complete version below if status labels or source details are missing.

Ask for help with fixing a result that still lacks source checks.

✔️ Awesome, I've got everything!

Your conversation now contains the complete source-checking prompt. Keep this result available for the source comparison next.

ⓧ I'd like to double check the full code

Compare your submitted Prompt 2 with this complete version:

Review the five items you just found.

For each item:
- Label it "Confirmed launch or announcement" only when the original company, product team, project, or researcher account posted the news.
- Otherwise label it "Rumour or hype."
- Name the source account and include a link to the original X post.
- If you cannot find an original source, write "No primary source found."
- Remove anything that falls outside the past 24 hours.

Keep the result short.

What should match?

The complete prompt defines two status labels. It also requires source accounts plus evidence links.

Its fallback phrase exposes missing primary evidence. Its time rule keeps the briefing current.

Compare one original post

A source link gives you a direct path from Grok's summary to the original claim. Opening one link reveals whether the summary preserves what the source actually said.

Choose one briefing item that includes a link to an original X post.

Before you open the link, do you expect the summary wording to match the original claim closely?

Open the original X-post link for your chosen item.
Identify whether the linked account is directly responsible for the news.
Compare the original post with Grok's short summary.

You should see the original account plus the wording of its claim. You can now judge whether Grok's status label reflects the available evidence.

Count Independent Sources

Repeated reposts can all trace back to one unsupported claim. The original responsible account provides stronger evidence than the number of times a story was repeated.

Source link not opening?

Return to the briefing and choose another item with an original X-post link.
Resubmit Prompt 2 if none of the listed links open.

Ask for help with checking a source link that does not support the summary.

Verify the source-aware briefing

A source-aware briefing shows the evidence standard beside every item. This final check confirms that weak or outdated stories cannot hide inside the list.

Before you inspect every item, do you expect all five original headlines to survive the evidence and time checks?

Review every remaining item in Grok's latest reply.
Confirm that each item uses either Confirmed launch or announcement or Rumour or hype.
Confirm that each item names its source account.
Confirm that each item includes an original X-post link or says No primary source found.
Confirm that every remaining item comes from the past 24 hours.

You should see a short briefing where every remaining story exposes its status plus its source evidence. Any item outside the requested time window should be gone.

That is the trust layer in place. Your scan now shows which headlines deserve attention before you act on them.

Your briefing can now separate stronger announcements from rumours and hype. Next, you will turn those credible updates into a personal briefing with relevance plus practical next actions.

Build Your Actionable Briefing

Your source-aware scan now separates confirmed announcements from rumours in Grok. You can inspect the evidence behind each surviving story.

Credible news still needs a personal filter before it can shape your day. A reusable prompt will connect each story to your interests before giving you one realistic next move.

In this step, get ready to:

Generate a personal briefing in a fresh Grok conversation.
Audit the result to remove unsupported claims or weak filler.
Save the completed reusable prompt for tomorrow.

Generate your personal briefing

A fresh conversation proves that Prompt 3 carries every instruction it needs. Its live X search starts from your watchlist without relying on the earlier scan.

Return to your written watchlist from earlier.
Use Grok's new-conversation control to start a blank chat.
Customize Prompt 3 by replacing each topic slot below with the exact wording from your watchlist:

Create my personal AI trend briefing for these topics: [[TOPIC_1="your first AI topic"]], [[TOPIC_2="your second AI topic"]], and [[TOPIC_3="your third AI topic"]].

Search live X posts from the past 24 hours and return the 3 to 5 most important items for those interests.

Prioritize original posts from companies, product teams, projects, and researchers. Separate the result into:
1. Confirmed launches and announcements
2. Rumours or hype

For each item, include:
- A short headline
- Its status
- The source account and a link to the original X post
- One sentence explaining why it matters to someone interested in my topics
- One practical thing I could try, test, or read next

Remove duplicates. If fewer than 3 trustworthy items exist, return fewer items rather than filling the briefing with weak stories. Keep the whole briefing concise and easy to scan.

What does this prompt do?

The past-24-hours window keeps the scan focused on recent activity.
The two status sections make uncertainty visible.
The source requirement prioritizes accounts directly responsible for each story.
The relevance line connects each item to your selected interests.
The practical action turns each useful update into something you can try.
The final rule permits fewer items when the available evidence is weak.

Delete the complete third-topic phrase from the first sentence if your watchlist contains only two topics.
Send your completed Prompt 3 in the fresh conversation.

You'll see a concise personal briefing split into confirmed news and hype. Each item should include its source plus a reason it matters to you.

Good progress. Your trend finder now turns a live scan into something shaped around your day.

Briefing missing a required field?

Check that you pasted the complete prompt through its final sentence.
Check that every topic slot contains one focused interest from your watchlist.
Keep the instruction that allows fewer items when trustworthy updates are limited.

If the structure still looks incomplete, help me check why my Grok briefing is missing required fields.

Audit the briefing

The first result is a draft built from fast-moving posts. Prompt 4 acts as a quality gate for recency, evidence, relevance, duplication, and realistic actions.

Audit the briefing by pasting Prompt 4 into the same fresh conversation:

Audit the briefing you just produced.

Check that every item:
- Comes from the past 24 hours
- Has the correct status
- Links to an original source when one exists
- Explains why it matters to my chosen interests
- Suggests one realistic next action

Remove duplicates, unsupported claims, and weak filler. Then return only the corrected briefing in 3 to 5 items, or fewer if there are not enough trustworthy updates.

What does this audit check?

The audit checks the time window for every item.
It checks whether each status matches the available evidence.
It looks for an original source when one exists.
It removes repeated stories or unsupported claims.
It keeps fewer items when stronger evidence is unavailable.

Before you send it, do you expect every item in the current briefing to survive the audit?

Send Prompt 4 to run the quality check.

Grok returns only the corrected briefing. Weak filler may disappear because the audit values trustworthy updates over a fixed item count.

Audit still leaving weak items?

Check whether the corrected result identifies an original source for each confirmed announcement.
Check whether unsupported items carry the status "Rumour or hype."
Send Prompt 4 once more if Grok explained the audit without returning the corrected briefing.

For help with a questionable result, help me evaluate whether this item belongs in my corrected briefing.

Save and verify your reusable prompt

Today's stories will change. Your customized Prompt 3 is the repeatable tool that preserves your topics plus your evidence standard.

Scroll back to your customized Prompt 3 in the fresh conversation.
Copy the entire customized Prompt 3.
Paste the prompt into the notes or document tool you use for reusable text.
Save the note or document.
Confirm the saved copy includes the opening topic line.
Confirm the saved copy includes the final instruction about concise output.

Your saved copy should show your selected topics at the top. Its final line should require a concise briefing that is easy to scan.

Before you inspect the final result, which requirement do you think is easiest for a weak item to miss?

Return to the corrected briefing in Grok.
Check that the briefing contains no more than five items.
Check that every included item has a short headline.
Check that every included item has a status.
Check that every included item names its source account.
Check that an original X-post link appears when one is available.
Check that every included item explains its personal relevance.
Check that every included item suggests one realistic next action.

You'll see three to five items when enough trustworthy updates exist. You may see fewer when the available evidence is limited.

Every surviving item should now have a status, a source, personal relevance, and one realistic action. Duplicates or unsupported claims should be gone.

You now have the repeatable tool. Tomorrow's stories can change while your evidence standard stays fixed.

Secret mission

Schedule Your Trend Finder

Your reusable prompt already produces the right briefing by hand. Schedule it as a daily Grok Automation when your account supports the feature without an upgrade. Test one run before locating its management controls.

Clean Up Your Resources

Choose whether to leave your manual trend finder ready for tomorrow, pause its optional schedule, or delete that schedule entirely. The core chat workflow uses the $0/month Free plan, so your completed manual workflow has no ongoing resource charge.

Resources you used:

An optional Grok Automation named Daily AI Trend Finder. This resource exists only if your account exposed Automations without an upgrade.

Keep everything running

No action is needed. Choose this if you want to keep running your trend finder manually or let the optional Automation produce its next scheduled briefing.

Keep your saved customized Prompt 3 available for tomorrow's manual run.
Leave the optional Automation active only if you created it without upgrading.
Inspect its next scheduled briefing at your selected time.

Pause - I'll come back to this later

Pause the optional schedule while keeping its instructions available for later. Your saved customized Prompt 3 remains ready for manual runs.

Skip this option if no Automation exists in your account.
Return to the Grok Automations area from earlier.
Select Daily AI Trend Finder from your Automations list.
Use its pause control.
Confirm the Automation shows a paused state.
Use its resume control when you want scheduled briefings again.

Delete - I don't want to use this again

Remove the optional Automation to prevent future runs. Your saved customized Prompt 3 keeps the manual trend finder ready to use.

Skip this option if no Automation exists in your account.
Return to the Grok Automations area from earlier.
Select Daily AI Trend Finder from your Automations list.
Use its delete control.
Confirm Daily AI Trend Finder no longer appears in the Automations area.
Keep your saved customized Prompt 3 for future manual briefings.

Nice Work!

That’s a wrap! You’ve built a personal AI trend finder in Grok that turns live X posts into a concise briefing you can inspect and act on.

What you learned:

Created a focused AI trend watchlist with two or three personal interests. You used live X search to find updates from the past 24 hours.
Applied source checking to trace stories back to original posts. You separated confirmed announcements from rumours and hype.
Saved a reusable trend-finder prompt that adds personal relevance and practical next actions. You also used a quality check to remove duplicates and unsupported claims.
Secret Mission: Checked whether Grok Automations were available without an upgrade. When available, you scheduled and tested the Daily AI Trend Finder as a daily briefing.

Ready to quiz yourself?