• AIDotDev Digest
  • Posts
  • AIDotDev Digest #007: lovable 2.0, Ethical AI Datasets, Deep Research for GitHub

AIDotDev Digest #007: lovable 2.0, Ethical AI Datasets, Deep Research for GitHub

Multiplayer Coding Arrives, Mozilla’s Ethical Tools Drop, and GitHub Gets Smarter

Inside today’s AIDotDev newsletter

AIDotDev is run by devs, for devs. We write what we’d want to read, fast updates, useful tools, and zero bullshit. We’re not trying to be the biggest, just the best for AI builders.

Thanks for reading

Sam from AIDotDev

AI isn’t just helping developers. It’s becoming one.

In this week’s AIDotDev, we’re diving into:

  • 2 new AI develop news for coding and creating

  • 3 learning resources to sharpen your coding

  • 3 new AI tools

  • 1 cool AI cool project to inspire you


Highlights:

  • Lovable 2.0 launches with multiplayer coding, a 10x smarter AI agent, and security scans for safer app building.

  • Mozilla’s Open-Source Toolkits empower ethical AI with tools for secure audio transcription and document conversion.

  • Project DeepWiki brings AI-powered deep research to GitHub, fueled by Devin.

Let’s build smarter. 🛠️

📰 AI DEV BITES

Lovable has launched version 2.0, bringing major upgrades to its AI-driven app and website creation platform with real-time collaboration, advanced AI agents, and robust security features.

The Details:

  • Multiplayer Collaboration: Lovable 2.0 introduces multiplayer mode, allowing teams to edit the same app in real time or collaborate across projects in shared workspaces—ideal for startups and enterprises.

  • Chat Mode Agent: The new chat mode agent is 10x smarter, designed for planning, debugging, and asking questions without editing code. It can reason across multiple steps, search files, inspect logs, and query databases.

  • Security Scan: A new security scanning tool, integrated with Supabase-enabled apps, identifies vulnerabilities before publishing. Results are available in the publish dialog, marking a step toward safer "vibe coding."

  • Dev Mode & Visual Edits: Dev Mode lets users directly edit code within Lovable, while enhanced visual editing tools allow for faster, more precise style adjustments—both features refined based on user feedback.

  • Custom Domains: Over 10,000 custom domains have been connected since the feature’s introduction, streamlining the process of buying and linking domains directly in Lovable.

  • Pricing Updates: Lovable simplified its pricing into two plans: Pro at $25/month and Teams at $30/month, offering flexible tiers for individual developers and collaborative teams.

  • New Brand & UI: A bold rebrand and cleaner UI reflect Lovable’s mission to empower users to turn dreams into reality, with a modern aesthetic featuring vibrant gradients and sleek design.

Developer Impact:
Lovable 2.0 transforms how developers and teams build apps, making collaboration seamless with multiplayer mode and empowering faster iteration through visual edits and Dev Mode. The smarter chat mode agent streamlines debugging and planning, while security scans help catch issues early, especially for Supabase users. However, as with any AI-driven tool, developers should still manually review code for edge cases, and the security features are just a starting point for more robust protections to come. These updates make Lovable a stronger choice for both prototyping and professional development, balancing creativity with practicality.

Mozilla has unveiled two open-source toolkits, developed in collaboration with EleutherAI, to support ethical AI development by helping developers avoid using copyrighted material in training datasets. These practical resources aim to empower the growing community focused on building high-quality, openly licensed alternatives to today’s opaque AI training methods.

The Details:

  • Speaches (Self-Hosted Audio Transcription): A local server using open-source Whisper models for private, secure audio transcription. Deployable via Docker or CLI, giving developers full control over sensitive data.

  • Docling (Document to Markdown Converter): A command-line tool to batch-convert PDFs, DOCX, HTML, and more into clean Markdown, with built-in OCR and image handling—perfect for creating structured datasets for AI training.

Developer Impact:
Speaches and Docling make ethical dataset creation practical and scalable. Backed by Mozilla’s AI Blueprints platform and new research into open dataset best practices, these tools empower developers to build trustworthy AI systems—though challenges like scaling and legal coordination remain.

📚 CURATED LEARNING RESOURCES

This video features a "60 Minutes" interview with Demis Hassabis, the 48-year-old British scientist who co-founded and leads Google's AI powerhouse DeepMind. In the interview, Hassabis discusses his work in artificial intelligence, including his Nobel Prize-winning research on protein structures, and shares his vision for the future of AI. The segment explores cutting-edge AI developments, including progress toward artificial general intelligence (AGI), demonstrates AI systems like Project Astra that can see and interpret the world, and examines both the potential benefits of AI (such as revolutionary advances in healthcare) and serious concerns about safety and control as these systems become increasingly powerful.

Learn how to build your own Voice AI Agent

I recently met Aemal Sayer at an event. He gave an excellent talk. His YouTube channel also covers a lot of great content. In this episode, he discusses using n8n and Vapi to create your own AI voice agent, similar to an "Own Voice" AI system.

Enhance your AI development

⚙️ NEW TOOLS

  • Bitnet.cpp - The official inference framework for 1-bit LLMs that support inference on CPUs.

  • Project DeepWiki - Deep Research for GitHub. Powered by Devin.

  • Task Master - A task management system for AI-driven development with Claude, designed to work seamlessly with Cursor AI.

🧸 AI-Powered Building: HOW PEOPLE DO IT

That’s a wrap for this week 👋

Got thoughts, tool tips, or cool AI projects we should feature? Just hit reply or drop us a line at [email protected] we read every message.

Until next time.

keep building the future!

Sam @ AIDotDev 🚀