Happy Sunday, all! Welcome back to another packed edition of the Roundup. This week, we're taking a look at the new Google search killer, OpenAI’s fair use legal woes, some of the most exciting and strange new launches, an exclusive story from a founder duo that pivoted 12 times, and more. Let's get into it. — Aaron and Sanjana  

Five of the most impactful or interesting launches we saw this week.

ChatGPT Search: Get fast, timely answers with links to relevant web sources

Arguably one of AI’s most important launches. ChatGPT Search is the AI company’s response to the likes of Perplexity and even Google. You can now use ChatGPT to browse the web for you, you can even set it as your default search engine. The clear benefit is the lack of ads when getting search results.

Claude for Desktop: Anthropic just released a desktop client for their flagship AI 

Not as groundbreaking as ChatGPT Search but still a good addition to an AI stack. The new desktop app does exactly what it says on the tin: it allows you to use the company’s flagship language model, Claude, on your desktop as opposed to going to the site.

Bolt.new: Prompt, run, edit & deploy full-stack web apps

Bolt is kind of a dream for building MVPs. Put in a simple prompt and watch it build a functioning web app ready to deploy with a single click. I tried it with “make a Product Hunt clone” and what do you know, within two minutes a fairly indistinguishable working clone was live on my screen.

Horse: A more organized browser 

Horse browser is perfect for people (like me) who need to declare tab bankruptcy constantly. It swaps tabs for Trails — nested groups of pages that capture the natural flow of each internet journey. That means no more tabs and most importantly trying to find the one you need in a sea of hundreds.

APIPark: An open-source AI gateway and API developer portal

APIPark lets you combine AI models and prompts into APIs — e.g. you could use OpenAI’s GPT-4 and custom prompts to create your own translation or data analysis API. The platform supports over 100+ AI models and provides real-time data to track your API usage.

Alright, buckle up: OpenAI’s just got dragged by one of their own. In a splashy interview with The New York Times, former researcher Suchir Balaji accused the AI giant of training ChatGPT on copyrighted content without permission, essentially treating copyright law like a minor suggestion. Balaji, who left in August, says the company's data-hoarding tactics could spell disaster for content creators and the whole online ecosystem. It’s a messy mix of ethics and legality, and Balaji isn't holding back.

OpenAI isn’t just sitting there, though. They’ve fired back, insisting that their data use falls under “fair use,” a defense they’ve clung to as lawsuits from authors and news organizations stack up. They claim it's all above board and critical for advancing AI. But Balaji has cranked the heat on an already boiling debate, casting doubt on whether OpenAI is playing fair or pushing boundaries for profit. As mentioned, it’s not the company’s first rodeo in copyright drama. This year, eight newspapers sued the AI company and Microsoft for copyright infringement.

Our take: First of all, copyright law needs an update. It’s not equipped to handle tech like AI. Second of all, copyright is important, but how newspapers go about it might need a fresh look. Instead of lawyering up, why not try a revenue-sharing model? That way, it’s a win-win or a lose-lose. Either way, it’s fair. 

This week, we’re bringing you a write-up from Vaibhav Gupta, co-founder of BoundaryML, about how he and his co-founder survived 12 hard pivots before settling on a viable idea. It’s a crazy, illuminating read. Vaibhav writes:

“The general theme of this entire timeframe can be summarized by us using our well-paying jobs as a safety net and attempting to throw spaghetti at the wall hoping it would stick. We were looking for signal BEFORE we committed.

Ideas we pursued:

• 6 months - An online coding bootcamp with Income-share-Agreements. Motivated by wanting to make the world a better place. 15 users, but I pivoted cause no one would pay.
• 6 months - Interactive Twitch ads (demo). Motivated by making $$ on ads and Aaron being a YouTuber in some past life. We had 0 users, did a lot of outbound and got 1 larger League of Legends streamer to want to use us. We pivoted after our W22 YC interview ended in 6 minutes… Michael Seibel (a YC partner and co-founder of Twitch) straight up asked “Do you guys even watch twitch?” Spoiler alert, we did not. I swear I’m still haunted by that question.”

Full article here.

Starting this week, we’ll periodically alternate our trends section with its converse — an ‘outliers’ section highlighting unusual launches we’ve recently seen on the site. These are a few products that stood out from the pack, either in problem statement or approach.

Laminar: An open-source, all-in-one platform for engineering AI products. It lets you trace your LLM app, run evaluations, label production data, and generate prompt improvements.

Stretch It: A minimalist, gesture-activated timer for your Mac menu bar. Just “stretch” (pull down) the icon and drop it to set a timer. It’s extremely simple and unintrusive.

Wand: An AI-powered design tool for iOS and Apple Pencil that lets you draw and edit anything. You can upload reference material and Wand will create a custom private style based on the images.

That’s all for this edition. Thanks for reading! Now go crush your week.