04 February 2025

OpenAI's Deep Research Tool DESTROYS Google Gemini (With Proof)

2Frameworks

7Insights

Frameworks in this episode

Strategy5 steps

Deep Research Tool Selection Scorecard

Choose a research tool by matching its strengths to the task

Read →

Productivity6 steps

Multi-Model Deep Research Workflow

Combine specialized AI researchers, then synthesize their strongest findings

Read →

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Myth Buster· 1

Myth Buster13:00

More Web Coverage Did Not Produce the Better Report

Gemini's report is useful and materially better than manual research, but the host finds its synthesis less detailed and contextual than OpenAI's. Generic observations about short-form video and scattered recommendations illustrate why broad discovery alone does not guarantee a strong executive report.

Gemini identifies valid trends but explains them shallowly
Its executive summary lacks comparable data and citations
Recommendations are sometimes grouped without a useful conceptual structure
OpenAI supplies more context about implications and execution
Source breadth and analytical quality are separate dimensions

“But if you look at the quality of the output, it's just not as good.”
Host · 13:30

“while Jim and I deep research is a very good product, and it's something I've been using a lot, it's nowhere near as thoughtful and…”
Host · 13:30

#gemini#research quality#synthesis#marketing

Hot Take· 1

Hot Take02:00

Sam Altman's Trillion-Dollar Claim for Research Agents

Sam Altman estimates that Deep Research could perform a single-digit percentage of economically valuable work, implying trillions of dollars in potential activity. The host calls this an extreme hype statement but connects it to a real market currently served by analysts and consulting firms.

Altman frames the product in terms of global economic work
Research and analysis represent a large professional-services market
The host distinguishes the underlying use case from the scale of the hype
OpenAI is positioning the agent as an alternative to costly research labor

“I think can do a single digit percentage of all economically valued agent can do a single digit percentage of all economic work in the…”
Sam Altman video · 02:00

“That is a crazy hype statement, an absurd hype statement”
Host · 02:30

#ai agents#economic impact#consulting#automation

Explainer· 3

Explainer01:00

What OpenAI Deep Research Does Differently

OpenAI Deep Research is presented as an agent capable of independently completing complex research tasks rather than returning an immediate chat response. It can browse and reason across websites, text, images, PDFs, academic reports, and case studies before producing a synthesized report.

Handles complex research tasks independently
Processes multiple information formats
Uses the o3 model for reasoning
Produces a report after an extended research run
Exposes aspects of its work while researching

“it can take very complex tasks and go and research them independently.”
Host · 01:00

“And it can browse the web, scan text, images, PDFs, much more, synthesize this reason through it and prepare a report for you.”
Sam Altman video · 01:30

#openai#deep research#ai agents#o3

Explainer11:30

Gemini Starts with a Research Plan Instead of Questions

Gemini handles research setup differently from OpenAI. Rather than asking a series of clarifying questions, it generates an editable research plan before starting, then makes the completed result easy to move into Google Docs for collaboration and sharing.

Gemini generates an editable research plan
OpenAI instead asks clarifying questions
Research duration appeared broadly similar in the test
Google Docs export supports collaboration
Google ecosystem integration is a notable advantage

“So instead of where open. Ai asks clarifying questions. Gemini Creates a detailed research plan.”
Host · 11:30

“the fact that it basically just creates a Google doc by default, and you can just open this a doc and share it and everything…”
Host · 12:30

#gemini#research planning#google docs#collaboration

Explainer15:30

OpenAI's 28 Sources Versus Gemini's 252 Websites

The side-by-side test reveals a large discovery gap: Gemini searches 252 websites while OpenAI uses 28 sources. The host attributes Gemini's breadth to Google's crawling and indexing strengths, while arguing that OpenAI's o3 model and access to varied formats produce more thoughtful synthesis.

Gemini searches 252 websites in the comparison
OpenAI uses 28 sources
Google excels at crawling and indexing web content
OpenAI considers formats beyond ordinary web pages
The o3 model appears stronger at reasoning over gathered evidence

“it searched 252 websites where open AI searched 28 websites, which is a massive difference in where you get that information.”
Host · 15:30

“252 websites is a ridiculous amount of context to search.”
Host · 16:00

#source coverage#web crawling#o3#gemini

Story· 1

Story06:30

Inside an Eight-Minute, 28-Source OpenAI Research Run

The host demonstrates OpenAI Deep Research on marketing strategy for a 25-person US technology company. The system asks clarifying questions, researches 28 sources for eight minutes, exposes activity logs and source links, and produces a detailed report with an executive summary, citations, trends, practices, challenges, and recommendations.

The system requests context before researching
The run lasts eight minutes across 28 sources
Users can inspect activity logs and source articles
The report includes citations and supporting statistics
The output resembles a professional analyst report

“it researched for eight minutes and 28 minutes through 28 sources.”
Host · 07:00

“This very much reads like McKinsey, professional analyst type of report with links and citations and stats that are validating the points that are coming.”
Host · 08:30

#openai#research demo#marketing strategy#citations

Takeaway· 1

Takeaway03:00

Why Dan Shipper Calls Deep Research a Curiosity Bazooka

Dan Shipper describes Deep Research through vivid transportation metaphors before calling it a “bazooka for the curious mind.” The host argues that this captures the central value of powerful AI research: dramatically expanding what a curious person can investigate and absorb.

Deep Research amplifies individual curiosity
The product can traverse a large body of human knowledge
Strong research tools compress days of report reading
The metaphor emphasizes leverage rather than simple convenience

“It's a chauffeur stretch limo for the information superhighway. It is a double decker bus tour, but you're the only passenger in the city. You're…”
Host quoting Dan Shipper · 03:00

“it's a bazooka for the curious mind.”
Host quoting Dan Shipper · 03:30

#curiosity#dan shipper#knowledge work#research