🛠️ AI Agents Weekly: Perplexity’s Deep Research, Tips for Building Agents, Using Reasoning Models, AI Agent Builder
Perplexity’s Deep Research, Tips for Building Agents, Using Reasoning Models, AI Agent Builder
In today’s issue:
Perplexity’s Deep Research is here
Tips for Building Agents
Best Practices for OpenAI o-series Models
Agentic AI Systems Applied to Finance
Postman has launched their new AI Agent Builder
DeepHermes 3 Preview combines normal LLM response modes with reasoning capabilities
Improving how AI agents work together, LlamaDeploy, and more.
Top Stories
Perplexity Deep Research Agent
Perplexity launches Deep Research, a new AI-powered research assistant that conducts comprehensive analysis by performing dozens of searches and reading hundreds of sources.
The tool is being made available for free to all users, with Pro subscribers getting 500 queries per day while non-subscribers receive a limited number of queries per day. Perplexit’s Deep Research is currently available on the Web, with iOS, Android, and Mac releases coming soon.
The system leverages iterative search and reasoning capabilities to create detailed reports across various domains including finance, marketing, technology, and health.
Perplexity reports that Deep Research achieves a 20.5% accuracy score on Humanity's Last Exam and 93.9% on the SimpleQA benchmark, outperforming other leading models like Gemini Thinking and o1. Most research tasks are completed in under 3 minutes, with results exportable to PDF or shareable as Perplexity Pages.
Tips for Building Agents
Anthropic recently released a blog post on how to build AI agents effectively. The authors sat down recently to discuss their ideas further. Below is a technical summary of the conversation along with the key technical points and recommendations:
Keep reading with a 7-day free trial
Subscribe to NLP Newsletter to keep reading this post and get 7 days of free access to the full post archives.