OpenAI Reveals Deep Research Agent

A Research Assistant Outperforms Human Analysis

Author

Mark Chepelyuk
February 03, 2025

Good afternoon,

đź“… On This Day: February 1, 1961 - The term “vaporware” is first used by Philip Elmer-DeWitt in a TIME magazine article. The term is now commonly used to describe software that has been long announced but hasn’t actually been released. At the time, many experts believed Microsoft was guilty of using vaporware announcements to keep customers from purchasing software from other companies (by convincing them that a Microsoft version was just around the corner)

Here’s what you need to know about AI today:

  • OpenAI launches Deep Research feature

  • o3-mini debuts with universal access

  • Altman admits open-source misstep

  • EU activates first phase of AI Act

  • Google X spins out AI agriculture startup

Today’s Deep Dive:

Image source: OpenAI

🔍 AI's Research Breakthrough

OpenAI's Deep Research feature autonomously conducts comprehensive research tasks in under 30 minutes.

Key Details:

  • Pro-tier exclusive launch

  • 100 queries per month

  • 26.6% on Humanity's Last Exam

  • 5-30 minute completion time

  • Specialized o3 version

  • Comprehensive citations

Why It Matters: When AI can do in 30 minutes what takes humans days, we're seeing a fundamental shift in knowledge work. This could transform how we approach research, analysis, and decision-making across industries.

Image Source: OpenAI

🧮 The Mini That Could

OpenAI’s o3-mini brings reasoning capabilities to all users, matching o1's performance at lower cost.

Key Details:

  • Free tier access

  • 63% cost reduction

  • 24% faster responses

  • Three reasoning levels

  • Technical domain focus

  • 150 daily messages for paid users

Why It Matters: With DeepSeek pushing the market, OpenAI's response shows how competition drives progress. By making these capabilities widely available while maintaining performance, they're expanding AI's reach.

Image Source: Reddit

🔓 Altman’s Open Source Admission

Sam Altman admits OpenAI has been "on the wrong side of history" regarding open source.

Key Details:

  • Internal disagreement noted

  • DeepSeek praise included

  • Future model timeline hints

  • New agents coming soon

  • Image generator teased

Why It Matters: When the leader of AI's most closed ecosystem admits they might be wrong about openness, it's not just an admission - it's a potential industry inflection point. Hopeful but most remain doubtful there will be meaningful change.

Image Source: EU

đź“ś Europe's AI Rules

The EU activates first phase of AI Act with hefty penalties for violations.

Key Details:

  • $35M euro maximum fines

  • "Unacceptably risky" systems banned

  • Immediate enforcement

  • Clear penalty structure

  • Regulatory framework active

  • Industry-wide impact

Why It Matters: The EU isn't just regulating AI - they're setting a global precedent. As the first major jurisdiction to implement comprehensive AI rules, they're effectively writing the playbook for global governance.

Image Source: Heritable Agriculture

🌾 The Agriculture Moonshot

: Google X launches Heritable Agriculture using AI for crop improvement.

Key Details:

  • Machine learning breeding

  • Yield improvement focus

  • Moonshot approach

  • Genome analysis

  • Accelerated development

  • Sustainability goals

Why It Matters: While everyone's focused on chatbots, Google's tackling food security with AI. It's a reminder that AI's biggest impacts might come from solving humanity's oldest challenges.

🛠️ Trending AI Tools

That's it for today! If you’re interested in sponsoring this newsletter email [email protected] to discuss sponsorship opportunities.