- CreateMode AI
- Posts
- How Deep Is Deep Research
How Deep Is Deep Research
A brief surface overview of deep AI tools
Kyle Richard Baxter
February 14, 2025
I have a PhD-level analysis in my pocket and yet I can’t figure out how to make the most of it...
A quick summary: Deep research is basically a super charged AI model that when prompted on a topic will go through a process of collecting and sourcing complex and deep data on the subject matter resulting in a comprehensive report that you can proclaim is the result of your profound knowledge and discovery on the topic.
And today the landscape of deep research tools has expanded with Perplexity AI's release of its Deep Research Alpha, to accompany OpenAI's ChatGPT Deep Research and Google's Gemini 1.5 Pro.
I used it to give us an overview:
Functional Comparison Matrix
ChatGPT Deep Research | Gemini 1.5 Pro Deep Research | Perplexity Deep Research Alpha | |
|---|---|---|---|
Processing Speed | 5-30 minutes | 15-45 minutes | 10-25 minutes |
Source Evaluation | Automated scoring | Manual adjustment | Multi-iteration verification |
Export Formats | CSV, JSON | Google Docs | PDF, Markdown |
Maximum Input Size | 50MB files | 10M token context | 25MB files |
Compliance Certifications | SOC2 Type 2 | ISO 27001 | GDPR compliant |
Pricing Model | $200/month (100Q) | $20/month (50Q) | Free (10Q) + $20/mo Unlimited |
How do I make the most of it?
It is impressive, but I find myself using deep research then having the urge to ask for a summary:
If I ask deep research to provide a summary is that technically just research? who is reading all of that stuff
— richrd (@richrdxyz)
6:30 AM • Feb 15, 2025
I have looked for some practical applications for inspiration and this one stood out:
A really cool example of Perplexity Deep Research: go through your portfolio, pull up information from the web about recent executive orders, and tell you which companies in your portfolio are impacted with a risk assessment. Something you would be paying a financial advisor 0.x%… x.com/i/web/status/1…
— Aravind Srinivas (@AravSrinivas)
5:15 AM • Feb 15, 2025
Along with this:
I spent 10 minutes prompting Deep Research with a complex tax scenario—telling it to find the optimal structure to maximize a handful of state & federal deductions that counteract each other.
It came back with a perfect model and found some things I never even considered.
If… x.com/i/web/status/1…
— Nikita Bier (@nikitabier)
4:50 PM • Feb 14, 2025
The implications and practical applications are endless. If only there was a pause button so that I could explore it fully before the next profound tool is released…