Daily AI Content Research System

Build a repeatable AI workflow to find high-ROI video topics daily.

Introduction

30 Second Summary

Coming up with fresh video ideas every single day is exhausting. You know what your audience wants to learn, but finding the exact topics that will get views and drive course enrollments feels like guessing in the dark.

In this project, you will build a complete AI-powered content research system that surfaces high-traffic video topics in under 15 minutes a day. You'll use Perplexity for live web research and Claude for scoring and calendar generation.

What You'll Build

A no-code daily research workflow that uses AI to find, score, and schedule video content across YouTube, Facebook, and TikTok for Datapot's data analytics training audience in Vietnam.

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

  • A set of reusable Perplexity research prompts that scan competitors, surface trending topics, and identify high-intent questions in your niche.
  • A Claude scoring framework that ranks every topic idea by search demand, competitor gaps, and course enrollment potential.
  • A structured weekly content calendar with platform-specific assignments, video hooks, and format recommendations.
  • Secret Mission: Add a performance feedback loop that uses last week's video metrics to refine next week's topic selection.

Are there any prerequisites?

None. This project uses only free-tier tools and includes setup instructions for everything you need. You do not need any coding experience or paid subscriptions.

Set Up Your Research Workspace

Before you can research video topics, you need the right tools in place. Perplexity handles live web research (competitor scans, trending topics, audience questions), while Claude synthesizes that research into scored, prioritized content plans.

In this step, you will set up both tools and create a content calendar template to store your final output.

In this step, get ready to:
  • Create a free Perplexity account and run a test search.
  • Verify Claude works in VS Code and create your project workspace.
  • Build a content calendar spreadsheet with the columns you need.
Create your free Perplexity account

Perplexity is an AI-powered search engine that pulls live data from the web and returns answers backed by cited sources. You will use it daily to discover trending topics and scan competitor content.

  • Open your browser and navigate to perplexity.ai.
  • Click Sign Up in the top right corner.
  • Create your account using your email, Google, or Apple sign-in.
  • Once signed in, you will land on the main search interface with an Ask anything... search bar in the center.

Run a test search to confirm everything works.

  • Type What are the most popular data analytics tools in 2026? into the search bar and press Enter.
  • Wait for Perplexity to return an answer with numbered source citations.

You should see a written response with blue citation links to the sources Perplexity used.

What is Focus in Perplexity?

Below the search bar, you will notice a Focus option that lets you narrow results to specific source types (All, Academic, YouTube, Reddit, and more).

In later steps, you will use the YouTube focus filter to find trending video topics specifically. For now, leave it on the default All setting.

What about Pro Search?

Perplexity offers a deeper research mode called Pro Search that pulls from more sources and reasons through complex questions. Free accounts get 5 Pro searches every 4 hours, which is plenty for your daily workflow.

You can always use unlimited basic searches for quick checks and save your Pro searches for the deeper competitor and trend analysis you will do in Steps 2 and 3.

Search not returning results?

Make sure you are signed in to your account. Anonymous searches work but may be limited. If you see an error, try refreshing the page or clearing your browser cache.

help me troubleshoot why Perplexity isn't returning search results

Verify Claude and create your workspace folder

Claude will be your synthesis engine. It takes raw research from Perplexity and transforms it into scored topic lists and structured content calendars.

  • Press Cmd+Space (macOS) or the Windows key (Windows) to open your search bar.
  • Type VS Code and press Enter to open it.
  • Open the Claude extension by clicking the Claude icon in the left sidebar (or pressing Cmd+Shift+P on macOS / Ctrl+Shift+P on Windows and typing Claude).
  • Start a new chat and type Hello, can you help me with content strategy? then press Enter.
  • Confirm you receive a response from Claude.

You should see Claude respond with a conversational reply in the chat panel.

Claude not responding?

Make sure the Claude extension is installed and enabled in VS Code. Check the Extensions panel (click the square icon in the left sidebar) and search for Claude to verify it shows as installed.

help me get the Claude extension working in VS Code

Now create your project workspace folder.

  • Open the integrated terminal in VS Code by pressing Ctrl+` (same on macOS and Windows).
  • Move to your Desktop by running this command:
cd ~/Desktop
  • Create your project folder by running this command:
mkdir datapot-content-research
  • Confirm the folder was created by running:
ls

You should see datapot-content-research listed in the output.

  • Open the new folder in VS Code by clicking File in the top menu bar.
  • Click Open Folder.
  • Navigate to your Desktop, select the datapot-content-research folder, and click Open.

Your VS Code Explorer sidebar should now show the empty datapot-content-research folder as your workspace root.

Folder not showing in VS Code?

Make sure you selected the datapot-content-research folder itself (not a file inside it) when clicking Open. The folder name should appear at the top of the Explorer sidebar on the left.

help me open my workspace folder correctly in VS Code

Set up your content calendar spreadsheet

Your content calendar is where the final output of your research system lives. Every scored topic, format assignment, and platform target gets recorded here for your content team to execute on.

Google Sheets

  • Open Google Sheets at sheets.google.com and sign in with your Google account.
  • Click Blank spreadsheet to create a new sheet.
  • Rename the spreadsheet by clicking Untitled spreadsheet in the top left corner and typing Datapot Content Calendar.
  • In Row 1, add the following column headers across cells A1 through I1:
  • A1: Date.
  • B1: Topic.
  • C1: Format (Long/Short).
  • D1: Platform (YouTube/Facebook/TikTok).
  • E1: Search Demand (High/Med/Low).
  • F1: Competitor Gap (Yes/No).
  • G1: Course Alignment.
  • H1: Hook Idea.
  • I1: Status (Researched/Scripted/Published).
  • Bold the entire header row by selecting Row 1 and pressing Cmd+B (macOS) or Ctrl+B (Windows).

Notion

  • Open Notion at notion.so and sign in (or create a free account if you don't have one).
  • In the left sidebar, click + New page.
  • Title the page Datapot Content Calendar.
  • Type /table and select Table - Inline to create a database table.
  • Add the following columns (click the + icon to the right of existing columns):
  • Date (Date type).
  • Topic (Text type).
  • Format (Long/Short) (Select type).
  • Platform (YouTube/Facebook/TikTok) (Select type).
  • Search Demand (High/Med/Low) (Select type).
  • Competitor Gap (Yes/No) (Checkbox type).
  • Course Alignment (Text type).
  • Hook Idea (Text type).
  • Status (Researched/Scripted/Published) (Select type).

What does each column track?

  • Format tells your team whether to produce a long-form tutorial (10-20 min), medium explainer (3-7 min), or short-form tip (30-60 sec).
  • Search Demand captures how many people are actively looking for this topic.
  • Competitor Gap flags topics that existing Vietnamese channels have not covered well.
  • Course Alignment maps each video to a specific Datapot course (Power BI, SQL, Python, etc.) so content drives enrollments.

Your calendar should now have 9 column headers with no data rows yet. You will populate it with scored topics in Step 4.

Not sure which tool to use?

Google Sheets works best if your team already uses Google Workspace. Notion works best if you want to add views, filters, and link content to other project pages later. Both are free and either works for this project.

help me decide between Google Sheets and Notion for my content calendar

Your research workspace is ready. Next up, you will build the Perplexity prompts that systematically scan your competitors and uncover content gaps in the Vietnamese data analytics space.

Build the Competitor Intelligence Prompts

Your research workspace is ready with Perplexity, Claude, and a content calendar template.

Now it's time to build the research engine that powers your content decisions. In this step, you'll create a set of reusable Perplexity prompts that systematically analyze what your Vietnamese data analytics competitors are doing on video platforms.

The goal is to turn competitor research from a vague "check what others are doing" task into a structured, repeatable intelligence-gathering process that takes minutes instead of hours.

In this step, get ready to:
  • Run a competitor content scan across five Vietnamese data analytics channels.
  • Identify content gaps that competitors are not covering.
  • Analyze winning video formats and hook patterns for data analytics education.
Build your competitor watch list and run the first scan

Before you can find content gaps, you need to know what already exists. A competitor scan shows you which topics are getting traction so you can spot patterns and opportunities.

  • Switch back to Perplexity in your browser.
  • Copy and paste the following prompt into the Perplexity search bar, replacing [competitor name] with your first competitor:
What video content has [competitor name] published recently on YouTube/TikTok about data analytics, Power BI, SQL, or Python in Vietnamese? What topics are getting the most engagement?
  • Run this prompt for each of your top 5 competitors, one at a time:
  • The Future Analyst (madzynguyen.com).
  • Mastering Data Analytics.
  • Thầy Sơn BI.
  • VEF Academy.
  • CoderSchool.

What does this prompt do?

This prompt asks Perplexity to search the live web for recent video content from each competitor. Perplexity pulls from YouTube, TikTok, and other indexed sources to show you what topics these channels are publishing.

By running it for all five competitors, you build a map of the content landscape in Vietnamese data analytics education. This map becomes the foundation for spotting gaps later.

  • After running all five searches, copy the key findings from each result into a single file.
  • In VS Code, create a new file in your datapot-content-research folder called competitor-scan.txt.
  • Paste all five competitor results into this file, with each competitor's name as a heading to keep them organized.
  • Save the file.

Use Perplexity's follow-up questions

After each search, Perplexity suggests follow-up questions. Use these to dig deeper into a competitor's most popular videos or recent uploads. You can ask follow-ups without using one of your 5 daily Pro searches.

Perplexity returning limited results?

Some competitors may have limited online presence. If a search returns very little, try adding the competitor's website URL or social media handle to give Perplexity more context. You can also try the YouTube Focus filter for video-specific results.

help me get better results from Perplexity for competitor research

Run a gap analysis to find underserved topics

Knowing what competitors publish is only half the picture. The real opportunity is in what they are NOT covering. These gaps represent topics where Datapot can own the conversation.

  • Start a new search in Perplexity by copying and pasting this gap analysis prompt:
In the Vietnamese data analytics education space on YouTube and TikTok, what topics related to Power BI, SQL, Python, data visualization, or career guidance are underserved? What questions do Vietnamese learners ask in comments and forums that existing channels haven't answered well?

What does this prompt do?

This prompt flips the perspective from "what exists" to "what's missing." Perplexity searches forums, comment sections, and community discussions to find questions that learners are asking but nobody is answering well on video.

These underserved topics are your highest-opportunity content. When you publish a video answering a question nobody else has covered, you face zero competition for that search traffic.

  • Review Perplexity's response and note the specific topics and unanswered questions it surfaces.
  • Create a new file in your datapot-content-research folder called gap-analysis.txt.
  • Paste the full gap analysis output into this file and save it.

You should now see two files in your workspace folder: competitor-scan.txt and gap-analysis.txt.

Results feel too generic?

If Perplexity returns broad answers about data analytics education globally rather than the Vietnamese market specifically, try adding "in Vietnam" or "Vietnamese learners" to your prompt. You can also use follow-up questions to narrow the focus to specific platforms like Vietnamese Facebook groups or local tech communities.

help me refine my gap analysis search for the Vietnamese market

Analyze winning video formats and hook patterns

Topics are only half the equation. The format you choose and the hook you open with determine whether anyone actually watches. This prompt researches what's working right now in data analytics education content.

  • Start a new search in Perplexity by copying and pasting this format analysis prompt:
What video formats (tutorials, career advice, tool comparisons, day-in-the-life, shorts/tips) are performing best for data analytics education channels on YouTube and TikTok in 2025-2026? What hook patterns get the highest engagement in the first 3 seconds?

What does this prompt do?

This prompt researches the structural patterns behind successful videos. It looks at what types of videos (tutorials vs career advice vs quick tips) perform best and what opening hooks grab viewers before they scroll away.

The 3-second hook insight is especially critical for TikTok and YouTube Shorts, where viewers decide instantly whether to keep watching. You'll use these patterns later when building your content calendar.

  • Review the results and note which formats and hook styles are highlighted as top performers.
  • Create a new file in your datapot-content-research folder called format-hook-analysis.txt.
  • Paste the format and hook analysis output into this file and save it.

Check your datapot-content-research folder in the VS Code Explorer sidebar. You should now see three files: competitor-scan.txt, gap-analysis.txt, and format-hook-analysis.txt.

Not seeing all three files?

Make sure you saved each file after pasting the results. In VS Code, check that you created all files inside the datapot-content-research folder and not in your root workspace. If a file ended up in the wrong location, drag it into the correct folder in the Explorer sidebar.

help me organize my research files in VS Code

Why save results as text files?

Saving your Perplexity outputs as files in your workspace means you can quickly paste them into Claude later without re-running the searches. It also builds a research archive you can reference across multiple content planning sessions.

You now have a solid intelligence baseline covering what competitors are doing, what they're missing, and which formats are winning. Next up, you'll build prompts that surface what your audience is actively searching for right now.

Build the Trend Discovery Prompts

Your competitor intelligence prompts give you a clear picture of what others are doing and where the gaps are. Now it's time to find out what your audience is actively searching for right now.

In this step, you'll build Perplexity prompts that surface trending topics, high-intent questions from people close to enrolling in a course, and breaking news from the Microsoft ecosystem. These three categories cover everything from daily content ideas to strategic conversion-focused topics.

In this step, get ready to:
  • Build a daily trend discovery prompt that surfaces what Vietnamese professionals are searching for right now.
  • Create a conversion-intent prompt targeting high-ROI topics from learners close to a purchase decision.
  • Build a Microsoft ecosystem newsjacking prompt that leverages Datapot's unique partner advantage.
Build your daily trend discovery prompt

Competitor analysis tells you what's already been done. Trend discovery tells you what your audience wants right now. These prompts use Perplexity's live search to find topics people are actively looking for today.

  • Switch back to Perplexity in your browser.
  • Click the Focus filter and select YouTube to narrow results to video content.
  • Paste the following trend discovery prompt and run the search:
What are Vietnamese professionals and students currently searching for about learning data analytics, Power BI, SQL, Python, or switching careers into data in 2026? Include trending questions, emerging tools, and seasonal topics (job hunting seasons, certification exam periods).

Why use the YouTube focus filter?

The YouTube focus filter tells Perplexity to prioritize results from YouTube. This means you see what people are actually watching and searching for on the platform where your long-form content lives.

Topics that already have search demand on YouTube are safer bets for your content calendar.

  • Review the results. You should see a list of trending topics, questions, and seasonal patterns relevant to Vietnamese data analytics learners.
  • Now switch the Focus filter back to All and run this second prompt to capture community discussions:
What data analytics topics are trending on Vietnamese tech communities, Facebook groups, and forums this week?

What does this second prompt add?

The first prompt finds what people search for. This second prompt finds what people talk about in communities. These often surface different types of topics. Community discussions reveal frustrations, debates, and emerging interests that haven't turned into formal searches yet.

  • Copy both sets of results into a new file called trend-discovery.txt in your datapot-content-research folder.
  • Confirm your file contains results from both the YouTube-focused search and the community trends search.

Getting vague or irrelevant results?

If Perplexity returns generic global results instead of Vietnamese-specific trends, try adding "in Vietnam" or "tiếng Việt" to your prompt. You can also try rephrasing to be more specific about the platforms ("on Vietnamese YouTube channels" or "in Vietnamese Facebook groups").

help me get better results from my Perplexity trend discovery prompt for Vietnamese data analytics content

Build your conversion-intent prompt

Not all topics are equal. Some attract viewers who are just browsing. Others attract people who are actively deciding whether to enroll in a course. These high-intent topics are your most valuable content opportunities because they reach people closest to a purchase decision.

  • Start a new search in Perplexity with the Focus filter set to All.
  • Paste the following conversion-intent prompt:
What questions do people ask right before deciding to enroll in a data analytics course? Think: 'Is Power BI worth learning in 2026?', 'How long to become a data analyst?', 'Best SQL course for beginners'. List 20 high-intent questions in Vietnamese data analytics education.

Why are these your highest-ROI topics?

Someone searching "Is Power BI worth learning in 2026?" is not casually browsing. They are actively weighing whether to invest time and money in a course. A video answering this question puts Datapot directly in front of someone ready to enroll.

These topics may get fewer total views than broad tutorials, but they drive far more course sign-ups per view.

  • Review the output. You should see a list of 20 questions that people ask when they're close to making an enrollment decision.
  • Copy the results into a new file called conversion-intent.txt in your datapot-content-research folder.

Perplexity returning fewer than 20 questions?

If you get fewer results than expected, use Perplexity's follow-up question feature. Type something like "Can you add more questions specifically about SQL, Python, and Microsoft Fabric courses?" to expand the list without starting a new search.

help me get a full list of 20 high-intent questions from my Perplexity conversion-intent prompt

Build your Microsoft ecosystem newsjacking prompt

Datapot is a Microsoft Training Services Partner. This means Datapot can authoritatively cover Microsoft product updates, certification changes, and official courseware in a way that no competitor can match.

Why newsjacking gives Datapot an edge

When Microsoft announces a new Power BI feature or changes a certification path, viewers want to hear from someone with official credentials. Datapot's Microsoft partnership makes these videos instantly more credible than the same content from an unaffiliated channel.

  • Start a new search in Perplexity with the Focus filter set to All.
  • Paste the following newsjacking prompt:
What are the latest updates, new features, or certification changes in Microsoft Power BI, Microsoft Fabric, Azure Data, or Microsoft certifications in 2026? Which changes would be most interesting to someone learning data analytics?

What should you look for in the results?

Focus on results that mention specific feature releases, deprecation announcements, new certification exams, or pricing changes. These are the updates that create urgency for viewers. Someone who just heard that a certification exam is changing wants to know what to do next.

  • Copy the results into a new file called newsjacking.txt in your datapot-content-research folder.
  • Verify your datapot-content-research folder now contains six files: competitor-scan.txt, gap-analysis.txt, format-hook-analysis.txt, trend-discovery.txt, conversion-intent.txt, and newsjacking.txt.

Perplexity showing outdated Microsoft news?

If results reference old announcements, try adding a time constraint to your prompt like "in the last 30 days" or "June-July 2026". Perplexity searches live web content, but sometimes older articles rank higher. Adding a date range helps filter for recent news only.

help me get fresh Microsoft ecosystem updates from my Perplexity newsjacking prompt

Save your prompts as templates

These three prompts form the research layer of your daily workflow. Consider saving them in a separate file called prompt-templates.txt so you can quickly copy-paste them each morning without rewriting.

You now have a complete research foundation. Your workspace has competitor intelligence, market gaps, trending topics, high-intent questions, and Microsoft news. Next up, you'll feed all of this into Claude to score and rank your topic ideas into a prioritized content plan.

Build the Claude Scoring Framework

You now have a library of raw research outputs covering competitor activity, content gaps, trending topics, and high-intent questions.

Raw research alone doesn't tell you which topics to prioritize. You need a scoring system that weighs each idea against Datapot's business goals. In this step, you'll build a Claude prompt that transforms messy research into a ranked, actionable topic list with platform assignments and calendar slots.

In this step, get ready to:
  • Build a Claude scoring prompt that evaluates topics on three dimensions.
  • Test the framework with your real Perplexity research outputs.
  • Create a calendar assignment prompt that distributes topics across platforms and days.
Build the scoring prompt in Claude

A good scoring framework prevents the common trap of chasing views without driving course enrollments. Your prompt will score every topic idea on three dimensions that map directly to Datapot's business: how many people are searching for it, whether competitors already cover it, and how naturally it leads to a course signup.

  • In your datapot-content-research workspace folder, create a new file called scoring-prompt.txt.
  • Copy the following scoring prompt template into scoring-prompt.txt:
You are a content strategist for Datapot, a data analytics training company in Vietnam (Microsoft Training Services Partner). Datapot offers courses in: Power BI, SQL, Python for Data Analytics, Power Automate and Power Apps, Microsoft Fabric, Business Analyst skills, and Data Analyst programs.

Your audience is Vietnamese professionals aged 22-35 looking to upskill or transition into data analytics roles.

Given the following research about trending topics, competitor gaps, and audience questions, score each topic idea on three dimensions (1-10 each):

SEARCH DEMAND (how many people are looking for this)
COMPETITOR GAP (how underserved is this topic by existing Vietnamese channels)
COURSE ALIGNMENT (how directly does this topic lead viewers toward enrolling in a Datapot course)

Calculate a TOTAL SCORE (sum of all three).

Then output the top 15 topics ranked by total score, with:
- Topic title (in Vietnamese)
- Suggested format (Long-form tutorial 10-20min / Short-form tip 30-60sec / Medium explainer 3-7min)
- Target platform (YouTube for long, TikTok for short, Facebook for medium)
- Hook idea (first line or visual)
- Which Datapot course it maps to

What do the three scoring dimensions measure?

Search Demand (1-10) measures how many people are actively looking for this topic. High scores go to topics that appear in YouTube autocomplete, Perplexity trends, and community discussions.

Competitor Gap (1-10) measures how underserved the topic is. A score of 10 means no Vietnamese channel covers it well. A score of 1 means multiple competitors already have strong videos on it.

Course Alignment (1-10) measures how directly the topic leads viewers toward enrolling in a Datapot course. A Power BI tutorial scores higher than a general career advice video because it connects to a specific product.

  • Save scoring-prompt.txt.

You now have a reusable scoring template that you can paste into Claude alongside any batch of research.

Test the scoring framework with your research outputs

The best way to validate a scoring prompt is to feed it real data and check whether the output makes intuitive sense. You'll paste your Perplexity research from Steps 2 and 3 into Claude alongside the scoring prompt.

  • Open the Claude extension in VS Code by clicking the Claude icon in the sidebar.
  • Start a new chat.
  • Paste your scoring prompt from scoring-prompt.txt into the chat.
  • Below the prompt, paste the contents of your research files: competitor-scan.txt, gap-analysis.txt, trend-discovery.txt, conversion-intent.txt, and newsjacking.txt.
  • Send the message and wait for Claude to return a scored list.

What should I see?

Claude should return a ranked table of 15 topics. Each topic should have a Vietnamese title, three individual scores (1-10), a total score, a suggested format, target platform, hook idea, and the Datapot course it maps to. The top-scoring topics should feel intuitively right. Topics with high search demand AND a competitor gap AND strong course alignment should be near the top.

  • Review the top 5 topics in Claude's output. Check whether they map to real Datapot courses like Power BI, SQL, or Python for Data Analytics.
  • If too many topics cluster around one course (e.g., all Power BI), or if the scores feel off, adjust the scoring weights. For example, to prioritize conversion over traffic, add this instruction to the end of your prompt: "Weight Course Alignment at 2x (multiply the Course Alignment score by 2 before calculating TOTAL SCORE)."
  • Update scoring-prompt.txt with any weight adjustments you made, then save the file.

Validate with YouTube autocomplete

For your highest-scoring topics, type the Vietnamese keyword into YouTube's search bar and check whether autocomplete suggests similar phrases. If it does, that confirms real search demand. If nothing appears, the Search Demand score might be too high.

Claude's output looks generic or ignores your research?

Make sure you pasted the actual research content below the scoring prompt, not just the file names. Claude needs the raw text to work with.

If the chat is too long and Claude seems to lose context, try sending the scoring prompt first, then paste research from only 2-3 files at a time and ask Claude to score based on that batch.

help me troubleshoot why Claude isn't using my research data to score topics

Create a calendar assignment prompt

A scored topic list is useful, but a content team needs to know exactly what to produce each day. This prompt takes Claude's ranked topics and distributes them across a 7-day calendar with platform-specific rules.

  • In your datapot-content-research folder, create a new file called calendar-prompt.txt.
  • Copy the following calendar assignment prompt into the file:
From the scored topic list above, assign topics to a 7-day calendar.

Rules:
- 1 long-form YouTube video per week
- 1 medium Facebook video every 2 days
- 1 short-form TikTok/Reels daily
- Avoid repeating the same tool (Power BI, SQL, Python) on consecutive days
- Include a mix of tutorial, career advice, and news/trend content

Output as a table with: Day, Platform, Format, Topic, Hook, Course Alignment.

Why these platform rules?

The cadence matches how each platform rewards content. YouTube favors fewer, longer videos with strong watch time. TikTok rewards daily posting of short hooks. Facebook performs best with medium-length explainers shared in groups and communities.

The "no repeat tools on consecutive days" rule ensures content variety. If Monday's TikTok is about Power BI, Tuesday's should cover SQL or career advice. This keeps the audience engaged across skill areas.

  • Save calendar-prompt.txt.
  • Return to your Claude chat (the one that already has the scored topic list).
  • Paste the calendar assignment prompt from calendar-prompt.txt and send the message.

What should I see?

Claude should return a 7-day table with one row per content piece. You should see a mix of platforms (YouTube, TikTok, Facebook) spread across the week. Each row should include the day, platform, format, a Vietnamese topic title, a hook idea, and which Datapot course it connects to.

A good calendar includes variety. Look for a mix of tutorials, career advice, and timely news content. If all seven days are Power BI tutorials, ask Claude to rebalance.

  • Verify that your calendar includes at least 1 long-form YouTube video, 3 medium Facebook videos, and 7 short TikTok pieces across the week.
  • Check that no two consecutive days feature the same tool (e.g., Power BI on Monday and Power BI again on Tuesday).

Calendar doesn't follow the platform rules?

If Claude ignores the frequency rules (e.g., assigns 3 YouTube videos), send a follow-up message: "Please rebalance. I need exactly 1 YouTube video for the whole week, 1 Facebook video every 2 days, and 1 TikTok daily. Redistribute topics accordingly."

help me get Claude to follow my content calendar distribution rules

You now have a complete research-to-calendar pipeline. Perplexity surfaces the raw opportunities and Claude transforms them into prioritized, platform-ready content plans. Next up, you'll assemble this into a repeatable daily workflow that takes under 15 minutes.

Assemble the Daily Workflow

Your scoring framework now transforms raw research into a prioritized, calendar-ready topic list. The hard intellectual work is done.

The final piece is packaging everything into a documented routine that anyone on Datapot's team can run in under 15 minutes. A system only has value if it runs consistently, so in this step you will document the workflow and execute your first complete cycle.

In this step, get ready to:
  • Document a repeatable daily workflow (10-15 minutes) and weekly workflow (20-30 minutes).
  • Run your first complete daily cycle from research to calendar entry.
  • Populate your first 7-day content calendar with scored, platform-assigned topics.
Document the daily workflow routine

A workflow that lives only in your head breaks the moment you hand it to a teammate or skip a day. Writing it down turns your research system into a team asset.

  • In VS Code, create a new file called daily-workflow.md inside your datapot-content-research folder.
  • Paste the following workflow document into the file:
# Datapot Daily Content Research Workflow

## Overview

Time required: 10-15 minutes
When to run: Every morning before content planning
Tools needed: Perplexity (free tier), Claude (VS Code extension), Content Calendar spreadsheet

## Step 1: Trend Discovery (2 min)

- Open Perplexity
- Run the trend discovery prompt from `trend-discovery.txt`
- Add today's date to the prompt for freshness (e.g., "as of July 2026")
- Use the YouTube focus filter for video-specific trends
- Copy the full output

## Step 2: Newsjacking Check (2 min)

- Run the Microsoft ecosystem newsjacking prompt from `newsjacking.txt`
- Look for: new Power BI features, Microsoft Fabric updates, certification changes
- Copy the full output

## Step 3: Claude Scoring (1 min)

- Open Claude in VS Code
- Paste the scoring prompt from `scoring-prompt.txt`
- Paste both Perplexity outputs (trend discovery + newsjacking) below the scoring prompt
- Send and wait for the scored topic list

## Step 4: Topic Review and Selection (3 min)

- Review Claude's top 15 scored topics
- Select 2-3 topics for today and tomorrow
- Prioritize topics with total scores above 24 (8+ average per dimension)
- Check for variety across platforms (YouTube/Facebook/TikTok)

## Step 5: Calendar Update (2 min)

- Add selected topics to the content calendar spreadsheet
- Fill in all columns: Date, Topic, Format, Platform, Search Demand, Competitor Gap, Course Alignment, Hook Idea
- Set Status to "Researched"

## Notes

- If a breaking Microsoft update appears, prioritize it regardless of score
- Rotate between Power BI, SQL, Python topics across consecutive days
- Keep Perplexity outputs saved for the weekly review session

What does this document do?

This workflow document breaks the entire research system into five timed steps that add up to 10-15 minutes. Each step references the exact prompt file to use, so a new team member can follow it without any context.

The "Notes" section captures decision rules you discovered during earlier steps. These prevent common mistakes like publishing three Power BI videos in a row or ignoring a major Microsoft announcement.

  • Save daily-workflow.md.
  • Confirm the file appears in your datapot-content-research folder in the VS Code Explorer sidebar alongside your eight existing files.

File not showing in the sidebar?

Make sure you saved the file inside the datapot-content-research folder, not in a parent directory. Check the file path shown in the VS Code tab title bar.

If the Explorer sidebar is collapsed, click the top icon in the left sidebar (two overlapping pages) to expand it.

Help me find where my file saved in VS Code.

Create the weekly workflow layer

The daily routine surfaces trending topics. But once a week, you need to step back and look at the bigger picture: what are competitors doing, and where is the full week heading?

  • Create a new file called weekly-workflow.md in your datapot-content-research folder.
  • Paste the following weekly workflow document:
# Datapot Weekly Content Research Workflow

## Overview

Time required: 20-30 minutes
When to run: Every Monday morning
Tools needed: Perplexity (free tier), Claude (VS Code extension), Content Calendar spreadsheet

## Purpose

The weekly session generates a full 7-day content calendar. Daily sessions then adjust for breaking news and trending topics discovered throughout the week.

## Step 1: Full Competitor Scan (10 min)

- Open Perplexity
- Run the competitor watch list prompt from `competitor-scan.txt` for each of the top 5 competitors:
  - The Future Analyst (madzynguyen.com)
  - Mastering Data Analytics
  - Thầy Sơn BI
  - VEF Academy
  - CoderSchool
- Run the gap analysis prompt from `gap-analysis.txt`
- Save all results

## Step 2: Conversion-Intent Research (5 min)

- Run the conversion-intent prompt from `conversion-intent.txt`
- These are the highest-ROI topics because viewers are close to a purchase decision
- Save results

## Step 3: Full Week Calendar Generation (10 min)

- Open Claude in VS Code
- Paste the scoring prompt from `scoring-prompt.txt`
- Paste all research from Steps 1-2 below the scoring prompt
- Include last week's calendar (copy from your spreadsheet) for context
- Send and review the scored topic list
- Then paste the calendar prompt from `calendar-prompt.txt`
- Review the 7-day calendar output

## Step 4: Populate the Calendar (5 min)

- Transfer the full 7-day calendar to your spreadsheet
- Fill in all columns for each entry
- Set all entries to Status: "Researched"
- Flag any topics that need additional research before scripting

## Calendar Rules (from calendar-prompt.txt)

- 1 long-form YouTube video per week (10-20 min tutorial)
- 1 medium Facebook video every 2 days (3-7 min explainer)
- 1 short-form TikTok daily (30-60 sec tip)
- No repeating the same tool (Power BI, SQL, Python) on consecutive days
- Mix of content types: tutorial, career advice, and news/trends

## After the Monday Session

- Daily workflow handles any adjustments for trending topics
- If a daily session reveals a high-score topic, swap it in for a lower-scoring planned topic
- Always keep at least 2 days of buffer content in case a day's research is thin

How do the daily and weekly workflows fit together?

The Monday session builds your full week's plan using deep competitor intelligence and conversion-intent research. This gives you strategic direction.

The daily session (Tuesday through Sunday) then layers in fresh trending topics and breaking news. If a daily discovery scores higher than a planned topic, you swap it in. This gives you tactical flexibility without losing strategic focus.

  • Save weekly-workflow.md.
  • Confirm your datapot-content-research folder now contains ten files: the eight from previous steps plus daily-workflow.md and weekly-workflow.md.

Only seeing nine files?

Double check that both daily-workflow.md and weekly-workflow.md are saved, not just open as unsaved tabs. Look for a dot or circle icon on the file tab indicating unsaved changes.

Help me check that all my workflow files are saved correctly.

✔️ Awesome, I've got everything!

Great. You now have two documented workflows ready to hand off. Next, you will run the full cycle for the first time.

ⓧ I'd like to double check the full files

Here is daily-workflow.md for comparison:

# Datapot Daily Content Research Workflow

## Overview

Time required: 10-15 minutes
When to run: Every morning before content planning
Tools needed: Perplexity (free tier), Claude (VS Code extension), Content Calendar spreadsheet

## Step 1: Trend Discovery (2 min)

- Open Perplexity
- Run the trend discovery prompt from `trend-discovery.txt`
- Add today's date to the prompt for freshness (e.g., "as of July 2026")
- Use the YouTube focus filter for video-specific trends
- Copy the full output

## Step 2: Newsjacking Check (2 min)

- Run the Microsoft ecosystem newsjacking prompt from `newsjacking.txt`
- Look for: new Power BI features, Microsoft Fabric updates, certification changes
- Copy the full output

## Step 3: Claude Scoring (1 min)

- Open Claude in VS Code
- Paste the scoring prompt from `scoring-prompt.txt`
- Paste both Perplexity outputs (trend discovery + newsjacking) below the scoring prompt
- Send and wait for the scored topic list

## Step 4: Topic Review and Selection (3 min)

- Review Claude's top 15 scored topics
- Select 2-3 topics for today and tomorrow
- Prioritize topics with total scores above 24 (8+ average per dimension)
- Check for variety across platforms (YouTube/Facebook/TikTok)

## Step 5: Calendar Update (2 min)

- Add selected topics to the content calendar spreadsheet
- Fill in all columns: Date, Topic, Format, Platform, Search Demand, Competitor Gap, Course Alignment, Hook Idea
- Set Status to "Researched"

## Notes

- If a breaking Microsoft update appears, prioritize it regardless of score
- Rotate between Power BI, SQL, Python topics across consecutive days
- Keep Perplexity outputs saved for the weekly review session

And here is weekly-workflow.md for comparison:

# Datapot Weekly Content Research Workflow

## Overview

Time required: 20-30 minutes
When to run: Every Monday morning
Tools needed: Perplexity (free tier), Claude (VS Code extension), Content Calendar spreadsheet

## Purpose

The weekly session generates a full 7-day content calendar. Daily sessions then adjust for breaking news and trending topics discovered throughout the week.

## Step 1: Full Competitor Scan (10 min)

- Open Perplexity
- Run the competitor watch list prompt from `competitor-scan.txt` for each of the top 5 competitors:
  - The Future Analyst (madzynguyen.com)
  - Mastering Data Analytics
  - Thầy Sơn BI
  - VEF Academy
  - CoderSchool
- Run the gap analysis prompt from `gap-analysis.txt`
- Save all results

## Step 2: Conversion-Intent Research (5 min)

- Run the conversion-intent prompt from `conversion-intent.txt`
- These are the highest-ROI topics because viewers are close to a purchase decision
- Save results

## Step 3: Full Week Calendar Generation (10 min)

- Open Claude in VS Code
- Paste the scoring prompt from `scoring-prompt.txt`
- Paste all research from Steps 1-2 below the scoring prompt
- Include last week's calendar (copy from your spreadsheet) for context
- Send and review the scored topic list
- Then paste the calendar prompt from `calendar-prompt.txt`
- Review the 7-day calendar output

## Step 4: Populate the Calendar (5 min)

- Transfer the full 7-day calendar to your spreadsheet
- Fill in all columns for each entry
- Set all entries to Status: "Researched"
- Flag any topics that need additional research before scripting

## Calendar Rules (from calendar-prompt.txt)

- 1 long-form YouTube video per week (10-20 min tutorial)
- 1 medium Facebook video every 2 days (3-7 min explainer)
- 1 short-form TikTok daily (30-60 sec tip)
- No repeating the same tool (Power BI, SQL, Python) on consecutive days
- Mix of content types: tutorial, career advice, and news/trends

## After the Monday Session

- Daily workflow handles any adjustments for trending topics
- If a daily session reveals a high-score topic, swap it in for a lower-scoring planned topic
- Always keep at least 2 days of buffer content in case a day's research is thin
Run your first complete daily cycle end-to-end

Documentation is only useful if the system actually works. Time to run the full workflow once and populate your first week's content calendar.

  • Open Perplexity in your browser.
  • Run your trend discovery prompt from trend-discovery.txt. Add today's date to the prompt for freshness (e.g., "as of July 2026").
  • Run your newsjacking prompt from newsjacking.txt to check for any recent Microsoft or data analytics news.
  • Copy both outputs from Perplexity.

Use the YouTube focus filter

When running the trend discovery prompt, select the YouTube focus filter at the top of the Perplexity interface. This narrows results to video-specific trends rather than general web content.

  • Switch to VS Code and open a new Claude chat.
  • Paste your scoring prompt from scoring-prompt.txt into the chat.
  • Paste both Perplexity outputs (trend discovery and newsjacking) below the scoring prompt.
  • Send the message and wait for Claude's scored topic list.
  • Review the scored output. Look for topics with total scores above 24 (strong across all three dimensions).
  • Now paste your calendar prompt from calendar-prompt.txt as a follow-up message in the same Claude chat.
  • Claude generates a full 7-day content calendar. Review it to confirm the output includes a mix of formats and platforms.

What should the calendar output look like?

Claude's output should show a table with 7 days of content. Each day has at least one entry, and the week includes:

  • 1 long-form YouTube tutorial (10-20 min). For example: "Hướng dẫn Power BI từ đầu cho người mới 2026".
  • Medium Facebook explainers every 2 days (3-7 min). For example: "Data Analyst vs Business Analyst: Khác nhau thế nào?".
  • Daily short TikTok tips (30-60 sec). For example: "3 hàm SQL bạn phải biết".

Each entry includes the topic, platform, format, hook idea, and which Datapot course it aligns to.

  • Open your content calendar spreadsheet.
  • Transfer Claude's 7-day calendar output into the spreadsheet. Fill in every column for each entry: Date, Topic, Format, Platform, Search Demand, Competitor Gap, Course Alignment, Hook Idea, and Status.
  • Set the Status column to "Researched" for every entry.
  • Verify that your calendar shows variety. No two consecutive days should feature the same tool (Power BI, SQL, or Python).
  • Confirm every entry maps to a specific Datapot course (Power BI, SQL, Python for Data Analytics, Power Automate and Power Apps, Microsoft Fabric, DBAP, or DAF).

Claude's output does not look like a calendar?

If Claude outputs a scored list but not a calendar table, make sure you sent the calendar prompt from calendar-prompt.txt as a follow-up message in the same chat. The calendar prompt needs the scored topics from the previous response as context.

If the calendar is missing platforms or has the wrong format distribution, remind Claude of the rules: 1 long-form YouTube per week, 1 medium Facebook every 2 days, 1 short TikTok daily.

Help me get Claude to output a proper 7-day content calendar.

Why run the full cycle now?

Running the complete workflow end-to-end reveals friction you cannot see on paper. Maybe the Perplexity output is too long and you need to summarize before pasting into Claude. Maybe the scoring weights need adjusting because too many topics cluster around the same score.

One real run teaches you more about the system than re-reading the documentation ten times. You will naturally refine the prompts after seeing actual output.

Secret mission

Add a Performance Feedback Loop

Your research system picks good topics, but how do you know which ones actually performed? Build a weekly review loop that feeds real video metrics back into your scoring framework, so the system gets smarter every week.

Clean Up Your Resources

Clean Up Your Resources

This project runs entirely on free-tier tools, so there are no ongoing costs to worry about. Decide whether to keep your resources for continued use, or delete them for a fresh start.

Resources you used:

  • datapot-content-research/ workspace folder (prompt templates, workflow documents, research outputs).
  • Content calendar spreadsheet (Google Sheets or Notion) with your 7-day calendar and Performance Log tab.
  • Perplexity free account (used for live web research).

Keep everything running

No action needed. Choose this if you plan to continue running your daily content research workflow.

  • Your datapot-content-research/ folder stays on your machine with all prompt templates and workflow documents intact.
  • Your content calendar spreadsheet remains accessible in Google Sheets or Notion.
  • Your Perplexity free account stays active for daily research sessions.
  • Your Claude VS Code extension continues working for scoring and calendar generation.

Pause - I'll come back to this later

Since all tools are free tier, there is nothing to shut down or pause. Your resources stay as they are with no cost.

  • Your local files, spreadsheet, and accounts will be waiting when you return.
  • When you come back, start with the daily workflow in daily-workflow.md to pick up where you left off.

Delete - I don't want to use this again

Remove all project files and resources to start fresh.

  • Delete your local workspace folder by running this command in your terminal:
rm -rf ~/Desktop/datapot-content-research

On Windows, run this instead:

rmdir /s /q %USERPROFILE%\Desktop\datapot-content-research
  • Delete your content calendar. In Google Sheets, open the spreadsheet, click File, then Move to trash. In Notion, open the page, click the ··· menu at the top right, then select Move to Trash.
  • Your Perplexity free account has no ongoing cost. If you want to remove it, log in to Perplexity, open your account settings, and look for the account deletion option.

That's a Wrap!

That's a Wrap!

Nice work! You've built a complete AI-powered content research system that Datapot's team can run daily in under 15 minutes.

You've learned how to:

  • Build a reusable prompt library using Perplexity that systematically researches competitors, trending topics, and high-conversion questions in the Vietnamese data analytics education market.
  • Develop a scoring framework in Claude that prioritizes video topics by search demand, competitor gaps, and course enrollment potential.
  • Assemble a documented daily workflow that produces a structured weekly content calendar with platform-specific assignments for YouTube, Facebook, and TikTok.
  • Secret Mission: Added a performance feedback loop that uses actual video metrics to refine scoring weights and improve future topic selection.

Ready to quiz yourself?