Opera's First AI Agentic Browser🌐

This week in AI innovation, the boundaries between humans and intelligent agents are dissolving faster than ever. From immersive voice interactions to browsers that act on your behalf, and powerful agent-building APIs—these updates signal a bold new chapter for task-oriented, human-in-the-loop computing:

• Opera unveils Neon, an alpha-stage, agentic browser reimagining how we surf the web—with tools like Chat, Do, and Make that let you research, automate, and create inside your browser.

• Mistral debuts its Agents API, giving developers the power to build persistent, context-aware AI agents that execute code, search the web, and collaborate across complex workflows.

• Claude rolls out voice conversations, transforming your mobile device into an AI-powered conversational assistant—with full voice input/output, context switching, and app integration.

Let’s dive into each breakthrough—and explore how they’re shaping the next era of intelligent software.

Opera Neon: AI-Powered Agentic Browser

Opera Neon is an innovative, agentic web browser developed by Opera, currently in its Alpha phase and available via invite-only waitlist. Designed for the emerging "agentic web," Neon integrates advanced AI features directly into the browsing experience. Its key tools—Chat, Do, and Make—allow users to get instant answers, automate web tasks, and generate content or applications without leaving the browser. Neon’s AI can understand user intent, interact with web pages, and perform actions such as research, form-filling, or even building web apps, all while prioritizing privacy and security. Opera Neon is subscription-based, with pricing details to be announced, and is aimed at users who want to shape the future of intelligent, task-oriented web browsing.

Mistral Agents API: Build Powerful AI Agents

Mistral AI has launched its new Agents API, a powerful tool designed to make AI more capable and practical by combining advanced language models with built-in connectors for code execution, web search, image generation, and document access, as well as persistent memory and agentic orchestration. This API enables the creation of intelligent agents that can perform complex tasks, maintain context across conversations, and coordinate multiple actions, making it ideal for enterprise-grade applications. Developers can build agents equipped with tools for secure code execution, dynamic image creation, document retrieval, and real-time web search, while leveraging the Model Context Protocol (MCP) for seamless integration with external systems. The API supports stateful, branching conversations and streaming outputs, allowing agents to collaborate and hand off tasks for efficient problem-solving in use cases like coding assistance, project management, financial analysis, travel planning, and nutrition advice. Mistral’s Agents API offers a robust framework for building agentic workflows, enabling developers to create, orchestrate, and deploy AI agents that can actively solve real-world problems.

Claude Now Gets Voice Conversations

Voice mode on Claude mobile apps is a beta feature available in English for both iOS and Android, allowing users to have full spoken conversations with Claude—speaking to the AI and hearing its responses, while key points are displayed on-screen. Users can easily switch between text and voice within the same conversation, with context maintained, and paid subscribers can access Google Docs, Calendar, Gmail, and web search through voice. To use voice mode, tap the sound wave icon in the app, select a voice, and start speaking; controls let you send, stop, or exit voice mode, and access files or the camera. Voice mode is ideal for hands-free planning, learning, creative thinking, interview prep, and capturing ideas, and is best used in quiet environments for optimal recognition. Usage limits depend on your subscription, with higher allowances for paid users. Voice transcripts are saved, and voice mode is designed with safety in mind, offering limited preset voices to prevent impersonation and enforcing all standard usage policies. Troubleshooting tips include ensuring a quiet environment, checking microphone permissions, and maintaining a stable internet connection.

Hand Picked Video

In this video, we'll look at building a thinking social media agent powered by Claude 3.7 Sonnet that can understand your voice, craft thoughtful posts, and engage authentically with your audience—all while saving you hours of content creation time.

Top AI Products from this week

• Clado - Clado puts 200 million+ people profiles (and growing) at your fingertips with simple, agentic search. If you’re interested in a global people search for sales, spreadsheets, hiring, and more, try our product.

• JoggAI 3.0 - Meet the next-generation AI ad tool that helps your product sell. Instantly turn your product into scroll-stopping photo and video ads with lifelike AI models that engage and convert like crazy.

• CodeRabbit VSCode Extension - Code, review, commit: all without leaving your IDE. CodeRabbit acts as a backstop that flags hallucination, logical errors, code smells, missed unit tests, and more.

• Coso.ai - Automatically creates engaging social media content for your brand each week, based on your brand’s personality and relevant social trends.

• Spoken Explore - Explore is a new way to browse for your home by vibe. Just click a piece you like, and we’ll show you more with the same feel. Powered by Spoken’s AI, which compares prices across 1,000+ stores by matching identical items—even when they’re renamed.

• Nomi - Listen to what people says on Zoom/Gmeet calls, and generate on-the-spot phrase suggestions, to help sales reps/teams increase win rates. You can also think of it as Cursor for Sales. Tab, tab, tab, and new suggestions pop.

This week in AI

• SignGemma - SignGemma, our advanced sign language-to-text model, joins the Gemma family soon—empowering inclusive communication for everyone!

• WordPress AI Team Launches - WordPress forms a dedicated AI Team to drive open, community-focused AI projects, speed innovation with plugins, and ensure strong integration across the ecosystem.

• DeepSeek R1 Update Released - DeepSeek’s updated R1 reasoning AI, now on Hugging Face under MIT license, offers a minor upgrade. At 685B parameters, it’s commercially usable but too large for most PCs.

• SpAItial Unveiled: 3D AI Revolution - SpAItial launches to build AI that creates and understands 3D worlds, pioneering Spatial Foundation Models for immersive media, robotics, and digital twins.

• Gemma 3n Preview - Gemma 3n debuts as a fast, efficient, multimodal AI model for phones and laptops—enabling private, on-device audio, text, and image understanding. Try the preview now!