Latest news with #GPT-4o
Fast Company
6 hours ago
- Business
- Fast Company
Beyond functionality: Building products people love to use
One of the most important realizations I've had in my time leading product and design at Rilla is that 'delight' is the heartbeat of a great product. No longer just a nice-to-have, delight is what separates a product that works from a product that people genuinely enjoy using. This is especially true in the era of AI, where functionality is rarely a differentiator. EVERY APP 'JUST WORKS' In 2025, AI tools have made functionality table stakes; every product more or less works. The bar for performance has been normalized. From AI code assistants like Cursor to design and prototyping tools like Figma Make, teams can build, design, and iterate at a pace that would've been unthinkable a few years ago. Transcription models like Whisper now produce near-perfect transcripts in real time. Chatbots built on GPT-4o or Claude can answer support queries, onboard users, or summarize meetings with almost no setup. Tasks that once took weeks now happen in days or even hours. Speed and capability are no longer bottlenecks. As a result, the differentiator isn't what a product does, but how it makes people feel while doing it. In this new landscape, delight isn't a bonus feature. It's the foundation of product love. This isn't a new idea. The most beloved products have always felt like someone cared deeply about every detail. Jony Ive, reflecting on Apple's design philosophy at the 2025 Stripe Sessions conference, said, 'I believe that when somebody unwrapped that box and took out that cable, and they thought, 'Somebody gave a shit about me,' I think that's a spiritual thing… It did genuinely come from a place of love, and from care.' In the past, that level of care sat on top of core functionality. Today, in a world where building things is becoming easier, it's what makes products stick. 1. Walk In Users' Shoes (Literally) In my experience, delight rarely comes from wireframes or metrics alone. Rather, it comes from feeling what users feel. That means looking past mocks and dashboards and into the messy, emotional, very human experience of actually using the product. This is why I encourage our engineers and designers to join support calls, build a personal customer advisory group to test ideas with, and even shadow users in person to truly understand their experience. You can't capture the sigh after the fourth loading spinner in a survey. You can't measure the hesitation before a confusing click. But by being close to your users, you can hear the catch in someone's voice as they hit a wall, or the quiet 'ohhh' when they finally get something to work. These are the moments that teach you why your product matters (or doesn't). 2. Delight As The Antidote To Bloat At Rilla, we ship fast, experiment often, and use AI where it helps. But in a world buzzing with AI features, it's tempting to pile on every new capability. Chasing delight helps us stay disciplined. It forces us to ask: Is this actually helping someone? If a feature adds complexity without clarity, it's noise. True delight is often rooted in simplicity—the kind that makes it instantly obvious what the goal is, and what happens when you get there. It shows up when every element has a purpose, when the interface feels obvious. When people don't just use the product, they get it. That kind of clarity builds trust. And trust is what makes people come back. 3. Turn Chores Into Moments Of Joy Some parts of a product are expected: forms, checklists, filters. They're not flashy. But they don't have to be painful. Delight lives in the details: a playful micro-interaction, a smart autofill, a helpful nudge that shows someone's thinking ahead for you. These touches turn a routine task into a moment of progress. The shift is subtle but powerful. It turns 'I have to do this' into 'That was smoother than I expected,' which turns into 'I kind of enjoy doing this.' When people feel capable, they feel good. And when a product makes someone feel good about themselves, not just the tool, that emotion is sticky. It's why they tell their teammates. It's why they come back. A NEW FRONTIER FOR CRAFT AND CONNECTION We're at a rare moment. Thanks to AI, the scaffolding is handled: the boilerplate, the routing logic, the endless setup. That means our time can shift from building what's functional to crafting what's meaningful for the people we're serving. The real opportunity in this new era isn't just speed. It's space. Space to care more. To notice more. To build with empathy and intentionality. When we obsess over the tiny moments—when someone smiles, breathes easier, or just feels understood—that's where product magic happens. Let's use this space well. Not just to build faster, but to connect deeper. To move beyond tools that 'work,' and toward experiences that feel personal, joyful, and unforgettable.
Tom's Guide
a day ago
- Business
- Tom's Guide
ChatGPT-5 launch expected soon — here's everything we know so far
OpenAI is reportedly gearing up to launch ChatGPT-5 in early to mid-August, and if the rumors are true, this next-gen model could be its most powerful update yet. According to reports from The Verge and Reuters, the release window is just weeks away (if not days), and it's set to bring major upgrades in reasoning, memory and multimodal abilities. Here's everything we know so far about ChatGPT-5, including what it can do, how it's different from GPT-4o and what to expect in terms of access and pricing. A post shared by ControlAI (@ai_ctrl) A photo posted by on That's the question on everyone's mind these days. While OpenAI hasn't confirmed an official launch date, sources close to the company say ChatGPT-5 is expected to debut in August 2025. That timing lines up with comments from OpenAI CEO Sam Altman, who recently hinted that 'GPT-5 is coming soon' and teased that it gave him a response so good, he felt 'useless compared to the AI.' Originally expected earlier this year, GPT-5 was reportedly delayed due to infrastructure constraints and further model tuning. But now, all signs point to a rollout happening within the next few weeks. ChatGPT-5 is OpenAI's next large language model, designed to unify several capabilities under one system. Get instant access to breaking news, the hottest reviews, great deals and helpful tips. Unlike GPT-4o, which combined voice, vision and reasoning in a more modular way, GPT-5 is said to integrate OpenAI's most advanced text and reasoning models, including the o3 series, into a single, smarter assistant. That means users can expect faster, more accurate responses across longer conversations, with better planning, fewer hallucinations and stronger contextual memory. Here's what we're expecting from ChatGPT-5: If OpenAI follows its current subscription structure, ChatGPT-5 will likely be available to ChatGPT Plus, Team, and Enterprise users at launch. That means you'll need to pay at least $20/month for early access. It's unclear whether free-tier users will get access to GPT-5 at launch or have to wait. There's also speculation that OpenAI's Pro tier will get the most advanced version of GPT-5, with faster speeds, access to Deep Research and advanced file tools. For everyday users, ChatGPT-5 could be a major upgrade, particularly if you rely on AI for writing help, idea generation, coding or productivity. If the new model delivers on its promise of better memory, logic and creative output, it could change how we interact with chatbots entirely. It also arrives at a time when Google's Gemini and Anthropic's Claude are rapidly improving, making the competition in consumer AI tools fiercer than ever. Follow Tom's Guide on Google News to get our up-to-date news, how-tos, and reviews in your feeds. Make sure to click the Follow button.
&w=3840&q=100)
Business Standard
4 days ago
- Business
- Business Standard
ChatGPT Agent arrives on Mac: Automates searches and tasks for subscribers
OpenAI's ChatGPT Agent feature launches on Mac for Pro, Plus, and Team users, enabling web browsing, data analysis, and task automation with GPT-4o and deep tool integration New Delhi OpenAI has reportedly rolled out its Agent feature to the ChatGPT app for Mac, offering AI-powered task automation for users on Pro, Plus, and Team plans. According to a report by 9To5Mac, the tool—announced a week ago—integrates OpenAI's Operator and Deep Research capabilities to enable hands-free, multistep task execution using GPT-4o. With this launch, users can delegate tasks such as browsing the web, creating presentations, conducting research, analysing data, and coding—directly from their Mac desktop. Availability What is ChatGPT Agent ChatGPT Agent is an AI automation assistant that can manage multistep workflows with minimal user input. Acting like a virtual operator, it can click, browse, type, and interact with web tools or platforms simulating human-like digital behaviour. On macOS, users can activate the feature using the '/agent' command or by selecting the new Agent button in the ChatGPT app interface. The agent follows a step-by-step reasoning process, visible to the user, who can step in or stop the process at any time. Key Capabilities In internal demos and user testing, ChatGPT Agent has demonstrated the ability to: Search and filter e-commerce websites, and add products to carts Draft presentations, pull data from Google Drive, and generate charts and slides Plan events or trips, find outfits, hotels, and populate spreadsheets Execute code and access terminal tools Conduct deep research and convert findings into structured formats Visually interact with websites via the Operator interface Limitations and Safeguards OpenAI emphasised that the Agent feature is still under development. It currently declines tasks involving financial transactions, legal matters, or sensitive communications unless explicitly approved by the user. Tasks like writing emails, placing orders, or editing shared documents require supervision or confirmation. Future updates will focus on enhancing speed, reliability, and safety. As OpenAI brings more powerful tools into desktop environments, ChatGPT Agent is poised to become a key productivity assistant for users relying on intelligent automation.
Yahoo
5 days ago
- Business
- Yahoo
OpenAI prepares August launch for GPT-5, The Verge reports
-- OpenAI is preparing to release its highly anticipated GPT-5 model as early as August, according to The Verge, citing sources familiar with the company's plans, marking a significant milestone in the evolution of generative AI. Internal testing and previously scheduled infrastructure upgrades fueled speculation earlier this year, with Microsoft Corporation (NASDAQ:MSFT) engineers said to be provisioning server capacity for a potential late-May launch. While delays have pushed the timeline slightly, OpenAI appears to be in the final stages of readying its most sophisticated model yet. GPT-5 has already surfaced behind the scenes, sharpening expectations around not only performance but also broader strategic implications for OpenAI's platform and API ecosystem. CEO Sam Altman confirmed the model's impending debut in a post on X and offered a firsthand account of its capabilities during a podcast appearance with Theo Von. Speaking about a question he had trouble answering, 'I put it in the model, this is GPT-5, and it answered it perfectly,' Altman said, describing the moment as leaving him feeling 'useless relative to the AI.' OpenAI plans to introduce mini and nano variants of GPT-5 alongside the core model, signaling a multi-tiered approach for developers and enterprises. These variants are designed to provide flexible options via API access, likely targeting a wide spectrum of applications across both consumer and commercial contexts. Unlike previous iteration rollouts such as GPT-4o and o3, GPT-5 has been hinted to unify OpenAI's leading capabilities, including multimodal comprehension, memory, and advanced reasoning, into a single model. Altman previously referred to it as 'a system that integrates a lot of our technology,' a framing that aligns with earlier comments about simplifying the user experience through fewer, more powerful versions. As the launch nears, investor and developer focus is shifting to how GPT-5 could streamline fragmented AI experiences and redefine expectations for general-purpose AI systems. With competition rising across the space, from Google's (NASDAQ:GOOGL) Gemini to Anthropic's Claude, GPT-5 is set to be OpenAI's bid to maintain its edge in the rapidly developing AI landscape. Related articles OpenAI prepares August launch for GPT-5, The Verge reports Surge of 50% since our AI selection, this chip giant still has great potential Apollo economist warns: AI bubble now bigger than 1990s tech mania Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
Tom's Guide
6 days ago
- Business
- Tom's Guide
Claude 4 vs ChatGPT: Which AI assistant is right for you?
Not all chatbots are created equal. OpenAI's ChatGPT and Anthropic's Claude are two of the most capable AI assistants available today, but they're built with different strengths, personalities and design philosophies. Whether you're looking for a creative collaborator, a coding partner or an all-in-one productivity tool, knowing how they compare can help you choose the right one for your style and needs. I've been testing AI chatbots for years, so let me break down the ChatGPT vs Claude differences that you need to know. While many users are eagerly anticipating the release of ChatGPT-5, OpenAI's current flagship model is GPT-4o (short for 'Omni'), and the company's most advanced base model to date. It supports multimodal input and output, meaning it can understand and respond to text, image, and audio, and it enables real-time interactions, including natural voice conversations and image generation. Anthropic's Claude 4, released in May 2025, comes in two versions. Claude Sonnet 4 is the free, fast, general-purpose model built for everyday use, while Claude Opus 4 is a premium model designed for deep reasoning, advanced coding tasks and large-scale context handling. It's built to power more complex workflows and long-form problem-solving. ChatGPT is available on the web, iOS and Android, and it's tightly integrated into Microsoft's ecosystem through Copilot in Bing, Word, Excel and other Office apps. Get instant access to breaking news, the hottest reviews, great deals and helpful tips. Claude 4, meanwhile, is accessible via on the web, iOS and Android. Developers and enterprise users can also access Claude via the Anthropic API, as well as through platforms like GitHub Copilot, AWS Bedrock and Google Cloud's Vertex AI. ChatGPT, powered by GPT-4o, offers strong conversational memory and supports multimodal input, including text, images, audio, file uploads and code execution. Users can personalize interactions with memory settings and even build custom GPTs for specific tasks. Although Claude does not have a memory feature, Claude 4 supports up to 200,000 tokens of context, allowing it to handle and retain significantly more information in a single conversation; the equivalent of more than 500 pages of text. ChatGPT offers fast, responsive performance and is particularly strong when paired with tools like Microsoft Copilot. Its reasoning abilities are enhanced by plugins, file uploads, and its new Agent feature, which can browse the web, execute tasks and automate workflows across apps. Claude Sonnet is similarly fast and excels at maintaining depth and continuity in conversations, making it a strong writing partner or analytical assistant. Claude Opus takes things further with support for multi-hour workflows, legal reasoning, code explanation and complex research tasks. On industry benchmarks like SWE-bench, Claude Opus ranks among the top performers for software engineering use cases. When it comes to multimodal support, ChatGPT offers more with in-chat image generation, visual context and real-time voice conversations with memory. Claude supports image input and visual understanding, meaning users can upload an image and the chatbot will assist with comprehension, but it cannot generate images. Also, Claude does not offer real-time conversational chat. Claude 4, especially in its Sonnet form, stands out for its natural, thoughtful tone. It often feels more like a helpful assistant offering clear explanations, thoughtful suggestions and a calm, conversational flow. If you're writing, brainstorming or trying to make sense of something complex, Claude tends to feel more human in how it is professional, polished and highly adaptable, with the ability to shift tones based on a user's preference or task. It is arguably more chatty than Claude, which appeals to some users more than others. ChatGPT supports a broad set of integrations including custom GPTs, an extensive plugin ecosystem, and Microsoft Copilot. Its new ChatGPT Agent feature (currently in early access) allows it to take real action on your behalf: browsing the web, interacting with forms, generating presentations and more all within a secure sandboxed environment. Claude 4 doesn't yet offer a native agent feature through its user interface, but it does support tool use and extended workflows through its API. It also integrates with GitHub Copilot, AWS Bedrock and other developer platforms, making it a strong option for those building their own agentic systems. If you're looking for an all-in-one AI assistant, ChatGPT is the more versatile choice. It can speak, see, listen, browse, generate images, analyze documents and automate tasks. Its voice mode feels remarkably human and is useful for a hands-free experience. The chatbot's ability to create and understand images adds another layer of functionality that Claude doesn't natively support. ChatGPT also integrates seamlessly with Microsoft products and lets users build custom GPTs tailored to their specific needs. And for power users, the new ChatGPT Agent adds another level of autonomy and actionability that's unmatched in most other consumer-facing AI tools. Claude 4 shines in natural language understanding and thoughtful conversation. It's ideal for users who prioritize clarity, warmth and coherence in writing, editing and reasoning. Whether you're crafting long-form content or analyzing dense documents, Claude delivers consistently thoughtful and easy-to-follow responses. Its 200K-token context window makes it the better choice for handling long inputs like research papers or legal contracts. Claude Opus, in particular, is built for high-stakes reasoning tasks and delivers top-tier performance for technical, legal and academic work. Even better, Claude Sonnet 4 is free, making it an excellent entry point for users who want fast, capable AI without paying for a subscription. Both ChatGPT and Claude are chatbots that I use every day. But they serve different goals. ChatGPT is my go-to for versatility, real-time interaction and multimodal tasks while Claude is what I use most often for summarizing, reasoning, deep context and analytics. Knowing what each excels at can help you choose the right assistant, or mix and match based on what you're trying to accomplish. Which chatbot do you prefer and why? Let me know in the comments. Follow Tom's Guide on Google News to get our up-to-date news, how-tos, and reviews in your feeds. Make sure to click the Follow button.
