We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
OpenAI Blog·Oct 23·5 min read
Consensus accelerates research with GPT-5 and Responses API
TL;DR
Researchers lacked efficient tools to synthesize evidence across scientific literature at scale. Consensus built a multi-agent research assistant using GPT-5 and OpenAI's Responses API—a framework for coordinating multiple AI agents—to automatically read, analyze, and synthesize research findings.
✦ Why It Matters
Engineers can leverage multi-agent architectures and modern language models to automate knowledge synthesis in specialized domains.
Key Takeaways
- Multi-agent AI systems can coordinate to solve complex tasks like literature synthesis at scale.
- Responses API enables orchestration of multiple AI agents working toward shared research goals.
- Automating evidence synthesis reduces researcher time investment from hours to minutes per analysis.
- Domain-specific AI assistants can achieve significant adoption when solving real bottlenecks in knowledge work.
Full Summary
Scientific researchers traditionally spend significant time manually reading and synthesizing evidence across multiple papers—a bottleneck slowing discovery. Consensus developed a multi-agent research assistant (multiple AI systems working together toward a shared goal) powered by GPT-5, OpenAI's latest language model, and the Responses API, a tool for orchestrating coordinated AI agent workflows.
The system automatically ingests scientific papers, extracts key findings, and synthesizes evidence across sources. By automating evidence synthesis, the platform reduces analysis time from hours to minutes while maintaining accuracy.
With over 8 million active researchers using the platform, Consensus demonstrates measurable adoption and impact on research velocity. This approach shows how agentic AI systems can scale domain-specific knowledge work beyond individual researcher capacity.