Yesterday's Top Launches: 5 Tools from July 15, 2026

Yesterday brought another wave of interesting tools into the world, with a particular focus on AI agents and practical utilities for developers and power users. If you’re keeping an eye on new developer tools that aim to simplify complex workflows, a couple of these launches are definitely worth a closer look.

Creatify

For anyone in marketing or running a small business, the pressure to create video content is relentless. Creatify steps in by automating the entire process of making video ads. You give it a product URL, and its AI gets to work generating a script, picking visuals, and even producing a voiceover. The idea of AI Ad Agents that can spit out a creative in minutes is compelling, especially for A/B testing campaigns on platforms like TikTok or Instagram.

While the promise is substantial, the real test will be in the quality of the output. Does the AI-generated script sound natural? Are the visuals genuinely engaging? For teams drowning in content demands, the freemium model makes it easy to try without commitment. It’s a tool that could save countless hours if it delivers on its core promise of creating high-converting ads from a simple link.

Osaurus

This one is for the privacy-conscious Mac user who’s wary of sending all their data to the cloud. Osaurus is a native macOS application that lets you run AI agents locally on your machine. Everything—your chats, the agent’s memory, its actions—stays on your computer. Built with Swift for Apple Silicon, it’s designed for performance and integrates with macOS features like Calendar and Contacts.

A standout feature is the approval gate; the agent has to ask for your permission before it takes any action, which adds a crucial layer of control and security. You can run models entirely offline using its Swift MLX runtime or bring your own API keys for cloud models like GPT-4, all while keeping your context local. It’s free and open source, which is fantastic, but be warned: you’ll want a Mac with plenty of RAM, ideally 24GB, to run larger local models effectively. For developers and tinkerers who want to explore agentic AI without the privacy trade-offs, Osaurus is a significant step forward.

Fabraix

As we deploy more AI agents, figuring out how they can be broken or manipulated becomes critical. Fabraix is a “frontier hacker” for AI agents, a platform designed to stress-test them by launching over a thousand adaptive attack strategies in real time. It operates as a pure black-box tester, meaning you just point it at your agent without needing to share any internal code.

The concept is solid. Traditional testing might not catch the weird, emergent ways an AI can fail. Fabraix proactively hunts for these vulnerabilities. There’s even a free “Playground” where you can try to find flaws in live agents, turning security research into a game. It’s open source and seems aimed at security researchers and teams building serious AI applications. For now, it feels like a niche but essential tool for anyone whose business depends on reliable and secure AI behavior.

Knockoff

Online shopping can be a minefield of weird, knockoff brands with names that seem generated by a keyboard smash. Knockoff is a simple but clever browser extension that cleans up your search results on big marketplaces. It uses a registered list of over 5,500 legitimate brands, a linguistic scanner to flag suspicious names (think too many consonants or random capitals), and a community reporting system.

The benefit is immediate: less clutter and a higher chance of buying from a reputable seller. It’s free, the data updates daily, and the code is open for inspection. For consumers, it’s a no-brainer. It doesn’t solve the broader issues of marketplace quality control, but it effectively gives you a filter for the noise, making the shopping experience noticeably better.

Marked QL

This is a classic utility app that solves a very specific pain point. If you work with Markdown files on a Mac, you’ve probably been frustrated that hitting the spacebar for a Quick Look preview just shows you the raw text. Marked QL fixes that. It’s a small application that adds a proper, rendered preview to macOS Finder, complete with syntax highlighting for code blocks, support for Mermaid diagrams, and MathJax equations.

Priced at $4.99, it’s a straightforward productivity boost for writers, developers, and anyone who lives in Markdown. It’s the kind of tool you install and then forget about because it just works, seamlessly integrating into your existing workflow. It’s hard to find a downside for its intended audience.

Community Favorites

Based on early user ratings from yesterday, two products quickly stood out. Osaurus generated significant buzz for its strong stance on privacy and local-first AI, resonating with developers. Marked QL was also highly rated for its immediate utility and polished execution, proving that a well-solved, focused problem is often more valuable than a broad platform.

Quick Links to Yesterday’s Launches: