TechCrunch Industry News Google launched its deepest AI research agent yet — on the same day OpenAI dropped GPT-5.2
Dec 12, 2025
Google's latest Deep Research tool, powered by Gemini 3 Pro, can now be integrated by developers into their own applications. It has promising use cases in areas like due diligence and drug research. The conversation also touches on the risks of AI hallucinations and efforts to mitigate them. Interestingly, this launch coincides with OpenAI's introduction of GPT-5.2, sparking competitive benchmarks between the two giants. Additionally, updates featuring encrypted chat and cryptocurrency capabilities were discussed, showcasing the evolving landscape of digital communication.
AI Snips
Chapters
Transcript
Episode notes
Google Embeds Deep Research In Apps
- Google released Gemini Deep Research to let developers embed powerful research agents in apps.
- The agent handles large context dumps and aims to minimize hallucinations during complex tasks.
Interactions API Targets Agentic Control
- Google's Interactions API gives developers more control for agentic AI integration.
- Minimizing hallucinations is critical because long agentic tasks amplify the impact of errors.
Google's New Benchmark And Results
- Google created DeepSearchQA to benchmark multi-step research agents and open-sourced it.
- On its benchmarks, Google's agent led while OpenAI's ChatGPT5 Pro was a close second.