Tech Files

ChatGPT Atlas is a new AI-powered web browser built with ChatGPT at its core. OpenAI views this development as a rare, once-a-decade opportunity to rethink what a browser can be and how people can use the web most productively and pleasantly.

In this episode of Tech Files, we explore OpenAI’s Sora 2, the upgraded generative AI video model that brings audio, continuity, and realism to AI-created clips. We break down how it’s shaping the future of content creation and dive into the ethical, technical, and creative implications of AI-driven video. Tune in as we unpack what Sora 2 means for creators, storytellers, and the media landscape ahead.

Orchids Cloner, a powerful new tool designed to replicate existing websites. Users can simply provide a link, and the AI-driven software will generate a "near pixel-perfect clone" of the original site. 

Discover GPT-5, OpenAI's latest major upgrade in AI intelligence, designed to be their smartest, fastest, and most useful model yet. It functions as a unified system that knows when to respond quickly and when to "think longer" for expert-level responses, effectively eliminating the need for users to choose between speed and depth.

OpenAI announces a significant advancement in AI, introducing a unified ChatGPT agent capable of performing complex, real-world tasks. This new agent integrates previous functionalities like Deep Research for in-depth internet searches and Operator for interacting with web interfaces, combining them into a single, more powerful tool. 

Autotab is a general AI agent that is designed to learn and work like a human. It is described as the first AI agent you onboard, not integrate. This means instead of complex technical integration, you teach it tasks

Lovable's new Agent Mode, a beta feature enabling the platform to autonomously think, plan, and execute actions similar to a human developer. This upgrade significantly reduces build errors and unwanted changes, helping users achieve their goals more effectively.

You can now work with Cursor Agents on web and mobile. Just like the familiar agent that works alongside you in the IDE, agents on web and mobile can write code, answer complex questions, and scaffold out your work.

In this episode, we explore Meta’s launch of its new AI app built on Llama 4, a bold first step toward more personal, voice-driven AI. Meta AI is evolving fast from WhatsApp to Instagram and now a standalone experience. Tune in as we unpack what this means for the future of human-AI interaction, how it fits into the broader AI race, and why this version might be the start of something much bigger.

Lovable 2.0, a significant update to the platform that allows users to build applications by chatting with AI. Key new features include a smarter Chat Mode Agent for planning and debugging, Multiplayer workspaces for team collaboration with different subscription levels, and a Security Scan feature to identify vulnerabilities. Additionally, the update introduces a new brand and UI, integrates custom domain purchasing, and presents simplified pricing options. 

Cluely is a desktop application that utilizes AI to secretly analyze a user's screen and audio during virtual interactions. This undetectable AI assistant aims to enhance user performance in areas like virtual meetings and sales calls by providing real-time assistance with information and responses.

The GPT 4.1 family of models, including GPT 4.1, Mini, and Nano, has been launched in the API, offering significant improvements over GPT-4o in areas such as coding, instruction following, and long context handling, now supporting up to one million tokens. These models are also more cost-effective, with GPT 4.1 being 26% cheaper than GPT-4o, and include the Nano model as the smallest, fastest, and cheapest option ever offered

Canva Code is a new feature designed to democratize coding. Canva Code allows users to create interactive experiences like quizzes and games without writing any code. By simply describing their desired outcome in a prompt, users can watch Canva AI generate the code.

In this episode of Tech Files, we dive into Firebase Studio—Google’s shiny new AI-powered workspace that might just change how apps get built forever. From importing repos and prototyping with natural language to using Gemini for code assistance, debugging, and documentation, Firebase Studio feels like a dev co-pilot on steroids.

Manus is presented as a general AI agent capable of executing diverse tasks ranging from research and analysis to planning and content creation.

Sync's latest innovation, lipsync-2. This advanced AI model offers zero-shot, style-preserving lip-synchronization for videos, meaning it can realistically match any audio to any person's mouth movements without prior training and while retaining their unique speaking style.

Runway Gen-4, a new AI model for image and video generation with world consistency, enabling consistent characters, objects, and environments across scenes for longer narratives.

Replit has released version two of its Agent tool. This updated version, Agent V2, boasts a tenfold increase in autonomy compared to its predecessor, enabling it to independently resolve problems. Users can observe their projects developing in real time within Replit's integrated environment, which supports the full development lifecycle from start to deployment.

Hedra Studio is presented as a next-generation multimodal content creation platform. It empowers users to produce engaging content by integrating generative image, video, and audio capabilities within a personal AI studio

OpenAI's new native image generation within the GPT-4o model in ChatGPT and Sora. This advancement aims to provide useful and precise image creation, moving beyond novelty by enabling accurate text rendering, adherence to detailed instructions, and learning from uploaded images. The "omniodel" architecture allows seamless integration across text, image, and audio modalities, fostering context-aware and consistent multi-turn generation.