Alibaba's Qwen Chat just launched its Deep Research feature for the general public following weeks of testing. Users provide a topic, answer a quick follow-up question to narrow the scope, and let the model generate a comprehensive report. Why it matters: Autonomous research agents are rapidly shifting from exclusive enterprise tools to standard, free chatbot capabilities.
Google is reportedly planning to unveil a Pinterest-like visual search feature at next week's I/O 2025 conference to defend its lucrative commercial search ads against ChatGPT. The company is also expected to showcase a software development AI agent alongside voice-driven Gemini integrations for Android XR and desktop Chrome.
Apple just had its recent AI shopping spree brought to light through EU regulatory filings, revealing the acquisition of four stealthy startups. The deals include smart scheduling app Mayday, assembly-line monitor Drishti, health data firm Betteromics, and RAG agent builder Pointable. Why it matters: Apple Intelligence is clearly reaching well beyond basic writing tools into factory floors and native apps.
OpenAI just launched its first connector for ChatGPT deep research, linking the tool directly to GitHub repositories. It can analyze codebases, map dependencies, and break down product specs into engineering tasks. Rollout starts with Plus, Pro, and Team users, with Enterprise and Edu following soon.
Microsoft is bringing support for the open Agent2Agent (A2A) protocol to Azure AI Foundry and Copilot Studio. It lets agents swap goals, manage state, and trigger cross-cloud workflows securely, so developers aren't locked into a single ecosystem. A Python sample is already live in Semantic Kernel.
Mistral just rolled out Mistral Medium 3, an efficiency-focused model claiming to match at least 90% of Claude 3.7 Sonnet's benchmark performance at $0.40 per million input tokens and $2 per million output tokens. Alongside the release, Mistral also made its corporate assistant, Le Chat Enterprise, generally available.
Anthropic just introduced native web search across its API and in Claude Code. Claude can now autonomously decide when to search, run multi-step queries, and return answers with source citations. It costs $10 per 1,000 searches plus standard tokens, saving developers from maintaining their own search pipelines.
Hugging Face just released Open Computer Agent, a free cloud-based tool that controls a Linux virtual machine to handle tasks. It's sluggish and easily tripped up by CAPTCHAs, but it shows open models are getting capable enough to power practical agent workflows.
FutureHouse just previewed Finch, an AI tool designed to accelerate data-driven biological research. The tool ingests papers, runs code, and produces scientific figures, acting essentially like a first-year grad student in a lab. It still makes occasional basic errors, and the team is currently onboarding computational biologists for closed beta testing.
ServiceNow and Nvidia just dropped Apriel Nemotron 15B, a compact open-source reasoning model built to run enterprise AI agents at lower latency and inference costs. Delivered as an Nvidia NIM microservice, it pairs with a new joint data flywheel architecture that feeds real enterprise workflow data back into model training.
Microsoft just unveiled a wave of new AI experiences for Windows 11 and Copilot+ PCs, headlined by an on-device agent in Settings that fixes PC issues via plain English. The update also brings dynamic photo relighting, text-to-sticker generation in Paint, and quick AI actions inside File Explorer. Microsoft is steadily weaving agentic automations directly into desktop plumbing.
Google just dropped an updated Gemini 2.5 Pro Preview ahead of I/O, packing big upgrades for coding and UI development. It took the #1 spot on the WebDev Arena leaderboard, handles complex backend refactors with ease, and can turn YouTube videos into interactive web apps. The update rolls out automatically to existing API endpoints at the same price.
Tether CEO Paolo Ardoino just announced Tether.AI, an open-source AI runtime aimed at connecting billions of AI agents across an unstoppable peer-to-peer network. The modular platform runs on any device without API keys, integrates peer-to-peer chat app Keet, and natively supports USDT and Bitcoin payments via Tether's Wallet Development Kit. Why it matters: Tether wants autonomous agents to trade and communicate without centralized choke points.
Google's Gemini 2.5 Pro has successfully beaten the 1996 classic Pokémon Blue after a run spanning hundreds of hours. An independent developer built an emulator harness that fed the model visual grids and memory stats while delegating tricky puzzles to sub-agents. It relied on custom scaffolding, but it's a telling milestone for how reasoning models handle massive, open-world planning.
Anthropic just launched Integrations, allowing Claude to connect directly to external tools like Jira, Linear, and Zapier via remote MCP servers. Claude also upgraded its Research mode to pull from connected workspace data and browse the web for up to 45 minutes to draft cited reports. Why it matters: Claude is shifting from a passive chatbot into an active workspace operator.
Amazon just made Nova Premier generally available in Bedrock, marking its most capable multimodal foundation model to date. It features a one million token context window for complex multistep reasoning, and functions as a teacher model to distill advanced performance into smaller, cheaper Nova models.
Vercept just launched Vy, an AI agent that runs natively on your computer, seeing your screen and directly using your signed-in apps. It is powered by VyUI, a new model that beats OpenAI, Google, and Anthropic on UI grounding benchmarks, and is available to download today.
Yelp just introduced an AI-powered call answering service to help restaurants and local businesses handle reservations, manage waitlists, and block spam calls. The tool can be tailored with custom greetings and routing rules so busy venues never miss a lead. It is currently under development.
Anthropic just published an Economic Index analysis of 500,000 developer interactions across Claude.ai and Claude Code. The specialized agent drove automation in 79% of sessions compared to 49% on the web chat, with front-end tasks dominating and startups adopting far faster than enterprises.
OpenAI just launched a lightweight version of Deep Research, expanding access to free tier users while boosting rate limits for Plus, Team, and Pro subscribers. Deep Research is easily one of the company's best agentic workflows, and making a leaner version means intense multi-step web research is finally hitting the masses.
