🤖 OpenAI launches ChatGPT Agent 🤖
Next-gen agent autonomously handles end-to-end workflow automation

OpenAI has introduced ChatGPT Agent, a unified agentic system that combines the capabilities of Operator's web browsing, Deep Research's analysis, and ChatGPT's conversational intelligence into a single platform.

Available to Pro, Plus, and Team users through agent mode in the tools dropdown, ChatGPT Agent can handle complex multi-step workflows like analyzing calendars, planning purchases, creating presentations, and conducting competitor analysis.

The system uses a virtual computer with multiple tools including visual browser, text-based browser, terminal, and API access, plus ChatGPT connectors for apps like Gmail and GitHub.

Unlike previous models, ChatGPT Agent supports iterative, collaborative workflows where users can interrupt, clarify instructions, or change tasks mid-execution without losing progress. The agent includes safety measures like tool limits, always-on classifiers for prompt injection, and approval requests for sensitive actions.

🔍 Google’s AI can make phone calls to local businesses for you 🔍
Gemini 2.5 Pro powers advanced research & automated phone calls

Google has launched Deep Search and AI-powered business calling features for Google Search, powered by Gemini 2.5 Pro model. Deep Search performs hundreds of searches to create comprehensive, fully-cited reports in minutes, ideal for in-depth research on jobs, hobbies, or major life decisions like home purchases. The AI business calling feature allows Search to automatically call local businesses to gather pricing and availability information on behalf of users, consolidating results from multiple establishments.

Users can search for services like "pet groomers near me" and select "Have AI check pricing" to trigger automated calls. Deep Search is available for Google AI Pro and AI Ultra subscribers in the US through the AI Mode experiment in Labs, while business calling is rolling out to all US Search users with higher limits for subscribers. Businesses maintain control through Business Profile settings.

🤖 Reflection AI unveils Asimov Code Research Agent 🤖
Multi-agent system focuses on codebase comprehension over generation

Reflection AI has launched Asimov, a code research agent designed for engineering teams that prioritizes codebase comprehension over code generation. The agent addresses the reality that engineering teams spend 70% of their time understanding code and only 10% writing it, yet most existing products focus primarily on generation. Asimov creates a single source of truth by ingesting entire codebases, architecture docs, GitHub threads, and chat history to build persistent memory of engineering systems. The platform features "Memories" that allow teams to capture tribal knowledge with senior engineers able to offload context via commands like "@asimov remember X works in Y way." 

Using a multi-agent architecture with long-context retrievers and a short-context reasoning agent, Asimov achieved 60-80% preference rates in blind testing with maintainers of major open source projects. The system includes Role-Based Access Control for knowledge management.

INTERESTING TO KNOW

💰 Anthropic launches Claude Financial Services Solution 💰

Anthropic has unveiled Claude for Financial Services, a comprehensive AI solution that transforms financial analysis by unifying market data and internal data sources into a single interface. The platform features Claude 4 models that outperform other frontier models in financial tasks, with Claude Opus 4 passing 5 out of 7 levels of the Financial Modeling World Cup and achieving 83% accuracy on complex Excel tasks. The solution includes pre-built MCP connectors for major data providers like FactSet, S&P Global, Morningstar, Databricks, and Snowflake, enabling real-time access to comprehensive financial information with direct hyperlinks to source materials.

🔧 IBM release API Agent for Intelligent API Management 🔧

IBM has announced the general availability of API Agent in IBM API Connect, an AI-powered assistant that transforms natural language prompts into fully-governed, production-ready APIs. Originally launched at IBM THINK in May and now out of public preview, the API Agent acts as an intelligent assistant trained on the API Connect platform and enterprise API estates, understanding both environment and intent. The agent autonomously plans and executes API management actions including discovering, creating, testing APIs, and supporting governance, while automatically deploying applications to free IBM Code Engine instances with no setup required.