Hey friends,
I’m creating a launch video for the interactive web app I shared last week (you got the first look in last week’s email 👀).
I needed to know: what actually makes SaaS product videos convert?
Real data from successful launches.
I could have spent hours reading case studies and blog posts.
Instead, I revisited ChatGPT Deep Research.
OpenAI just upgraded it with GPT 5.2 earlier this month.
20 minutes later, I was reading one of the most insightful reports I’ve read.
Like it was so pin point on what I needed to knw (AI likes to waffle).
So this update was a solid upgrade.
Let me show you what's changed and why I'm using this way more now.
📌 What Is OpenAI's Deep Research?
Deep Research is OpenAI’s dedicated research mode inside ChatGPT.
You give it a task, it spends 5-20 minutes analyzing hundreds of sources, and returns a structured report with citations.
The February 2026 update upgraded the agent model behind it to GPT 5.2 and that’s what makes the difference.
Here’s what actually changed:
- Domain-specific research: You can now restrict searches to trusted sources
- Real-time interrupting: Watch queries happen live and interrupt mid-research if it’s going the wrong direction
- Way less waffling: 30% fewer errors, tighter writing, and it gets to the point faster
✨ How I Actually Used It
✅ 1. Marketing
I’m creating a public launch video for the web app I shared with you last week.
You had the first chance to test it (now I’m prepping for wider release).
I needed data on what hooks viewers in the first 5 seconds, optimal video length, and examples from launches that actually converted.
My prompt: “Research best practices for SaaS product launch videos. Focus on: optimal length for conversion, what hooks viewers in the first 5 seconds, and examples from successful launches.”
20 minutes later:
- Specific hook strategies from real launches (not generic advice)
- Data showing 45-90 seconds converts best for launch videos
- Platform-specific recommendations with conversion benchmarks
- Three case studies with before/after metrics
I used those insights to plan my video.
I’m still working on that video btw.
Pro tip: Restricting to trusted sources helps you avoid generic advice and get strategic insights from people you trust.
✅ 2. Data-driven Insights
I’ve been using AI coding agents more lately and wanted to understand which ones people actually use.
Prompt: “Research the most popular AI coding agents. Focus on: usage patterns, key features, and what developers prefer. Research: Claude Code, Codex, Cursor, Replit Agent, and GitHub Copilot.”
18 minutes later: a comprehensive breakdown showing what makes each one different, which ones developers prefer for specific tasks, and detailed feature comparisons.
18 minutes for AI to do a task sounds long.
I don’t know about you… but I would have taken hours doing that type of research from multiple articles and forums.
✅ 3. Competitor Analysis
I was curious: what types of content do successful AI newsletters share that their audiences love?
Prompt: “Analyze content strategy for top newsletters. Focus on: types of content that get the most engagement, format patterns, and what makes their audience engage. Research: Ben’s Bites, TLDR AI, Superhuman, and Matt’s Future Tools.”
15 minutes later: a comparative analysis showing what content formats drive opens, what topics get replies, and patterns across different newsletter styles.
I spotted content gaps I wouldn’t have seen reading them individually.
This helps me understand what’s working in the space without manually tracking it all.
Pro tip: Use this for competitive analysis when you want to understand patterns fast. It’s great for initial analysis before you dive deeper into specific areas.
🛠 How to Try It (3 Steps)
- Access Deep Research: If you have ChatGPT Plus, Team, Pro, or Enterprise, you’ll see “Deep Research” when starting a new chat*
- Write a specific, focused question: Instead of “research new design tools” try something like “Compare the top 5 AI-powered research and design tools for solopreneurs in 2025, focusing on ease of use, collaboration features, and value for money.
- Restrict sources to what you trust: When you start a Deep Research query, you can select specific trusted sources.
*Free users also have access but may get a different model with fewer queries per month.
📈 Why This Actually Matters
Before I start any major project now, I run a Deep Research query.
Not only does it help me feel like a scientist 🧐
But because it helps me understand whether I invest more time and go down that path or not (especially for complex projects or problems I’m trying to solve).
I don’t need to spend hours reading, cross-referencing, and synthesizing anymore.
Deep Research handles finding the right sources, pulling out key insights, and showing me patterns I might miss on my own.
For creators, founders, and anyone doing knowledge work: this reduces the “gathering” phase so you can get to the “building” phase faster.
🫡 Final Thoughts
I thought Gemini Deep Research was forever the best, and it’s still really good.
But ChatGPT Deep Research with the GPT 5.2 upgrade gives it a solid competitor now.
Here’s what I use it for: understanding market landscapes before launching something new, researching strategies that have worked for others, and getting credible data fast when I need to make decisions.
It’s still slow (5-20 minutes per query).
But when you need strategic research (not quick facts), it’s worth the wait.
If you’ve got access, test it this week on a real project.
Don’t just try a demo question.
Give it actual work and see if it changes how you approach research like it did for me.
And if you’re curious about that launch video I keep mentioning.. it’s coming very soon 👀
To better research and smarter decisions,
Nahid
