• unwind ai
  • Posts
  • New AI Software Engineer Surpasses Devin

New AI Software Engineer Surpasses Devin

PLUS: In-browser AI Databases, Finetune your AI Agents

Today’s top AI Highlights:

  1. Cosine releases AI software engineer that surpasses Devin and Amazon’s AI

  2. Postgres.new - instant in-browser databases, supercharged with AI

  3. Eleven Labs releases fully managed AI dubbing studio

  4. Bet on events, get real-time event summaries with Perplexity AI

  5. Finetune your AI agents with synthetic users

& so much more!

Read time: 3 mins

Latest Developments

AI startup Cosine, focusing on understanding human reasoning, has released Genie, a state-of-the-art, fully autonomous AI software engineering colleague. Genie boasts the highest score of 30.08% on SWE-Bench, surpassing Amazon’s Q Developer and Congnition’s Devin. Cosine captured the cognitive processes of human software engineers in Genie's training data. This involves the intricate steps of reasoning, problem breakdown, and decision-making that human engineers follow in real-world development tasks.

Key Highlights:

  1. Long context - Genie leverages a long context window model, allowing it to access and retain a vast amount of information for sophisticated problem-solving capabilities.

  2. Human-like Reasoning - Unlike other AI-powered tools that rely on prompting-based models, Genie is trained on a comprehensive dataset that reflects the logical workflow of human engineers.

  3. Debugging Capabilities - Genie incorporates debugging tools to analyze application state and execution flow, further mimicking the approach of human developers.

  4. Iterative Problem-Solving - Genie can iterate through different approaches to problem-solving, learn from mistakes, and continuously refine its solutions, much like a human engineer.

You can apply for Genie’s access here.

Get the power of Postgres and an AI assistant right in your browser with postgres.new. This tool lets you spin up unlimited, free Postgres databases instantly, directly in your browser, powered by the magic of WebAssembly. It's perfect when you want to experiment with data, prototype quickly, or just want to learn SQL in a fun and engaging environment. And the best part? It's supercharged with AI for effortless data manipulation and visualization.

Key Highlights:

  1. AI-Driven SQL and More - The built-in AI assistant can write SQL queries for you, import and export CSVs, generate insightful charts, and even build out your database schema complete with ER diagrams and migration scripts.

  2. Instant Database Creation - No more waiting for downloads or complex setups. postgres.new lets you create and discard databases as quickly as you need them so you test and experiment without any overhead.

  3. Data Visualization - Generate beautiful charts in seconds by simply adding the word "chart" to your AI instructions.

  4. Powered by PGlite and WASM - Experience the speed and efficiency of Postgres compiled to WebAssembly with PGlite. This allows postgres.new to run entirely in your browser, making it incredibly fast and resource-light.

Quick Bites

  1. Eleven Labs has launched ElevenStudios, a fully managed dubbing platform that translates and dubs video and podcast content into multiple languages using AI and expert dubbers. They are live with some of the top content creators like 20VC and Colin & Samir, giving them high-quality dubbing while maintaining their unique voice and tone.

  1. Meta and Universal Music Group expanded their licensing deal to combat unauthorized AI-generated content and protect artists’ rights. The agreement allows sharing UMG music across Meta platforms, including WhatsApp and Threads.

  2. Perplexity AI has partnered with prediction marketplace Polymarket which lets users bet on real-world events. Now, when you search for events on Perplexity, you'll see news summaries paired with real-time probability predictions, such as election outcomes, market trends, and beyond.

Tools of the Trade

  1. Finetune: Refine your AI agents by testing them against synthetic users and generating performance reports and execution graphs. These refined agents can be easily deployed to a secure Virtual Private Cloud.

  2. AI Forms by Taskade: Create shareable and customizable workflows using automation. You can build forms to collect data and use automation to process that data with AI. When your form is processed, the form’s output will appear on the same page.

  3. LibreChat: Opensource, self-hosted ChatGPT clone that integrates multiple AI models and offers advanced features like conversation branching, multimodal chat, and customizable presets. 

  4. Awesome LLM Apps: Build awesome LLM apps using RAG for interacting with data sources like GitHub, Gmail, PDFs, and YouTube videos through simple texts. These apps will let you retrieve information, engage in chat, and extract insights directly from content on these platforms.

Hot Takes

  1. Asked someone using llm chatbots in production: “how do you know when the model is giving wrong answers?”, the response? “when the user starts complaining” lmao. get some evals in man ~
    anton

  2. The B2B & consumer startup ecosystems of today grew up in an era of stable technological change. Startups have strategies (like lean) that work well for stable environments. Funders have rules-of-thumb for stable tech.
    If AI growth is fast, it may be they are too unimaginative. ~
    Ethan Mollick

Meme of the Day

That’s all for today! See you tomorrow with more such AI-filled content.

Real-time AI Updates 🚨

⚡️ Follow me on Twitter @Saboo_Shubham for lightning-fast AI updates and never miss what’s trending!

Unwind AI - Twitter | LinkedIn | Instagram | Facebook

PS: We curate this AI newsletter every day for FREE, your support is what keeps us going. If you find value in what you read, share it with at least one (or 20) of your friends!

Reply

or to participate.