- unwind ai
- Posts
- Last Week in AI - A Weekly Unwind
Last Week in AI - A Weekly Unwind
From 4-Aug-2024 to 10-Aug-2024
Shubham Saboo & Gargi Gupta
August 11, 2024
It was yet another thrilling week in the AI field with advancements that further extend the limits of what can be achieved with AI.
Here are 10 AI breakthroughs that you can’t afford to miss 🧵👇
LangChain has released LangGraph Studio, the first dedicated IDE for AI agent development, in open beta. It offers a visual and interactive way to build, debug, and understand complex LLM-powered applications built with the LangGraph framework. It lets you see your agent's decision-making process in action, tweak its behavior on the fly, and even change the code underneath – all in real-time.
OpenAI is seeing a row of key departures. John Schulman, a co-founder of OpenAI, has left the company to join rival AI startup Anthropic. At the same time, OpenAI president Greg Brockman announced he will be taking an extended leave until the end of the year.
Schulman has joined Anthropic to “deepen his work in AI alignment”. He stated, “I believe I can gain new perspectives and do research alongside people deeply engaged with the topics I’m most interested in.”
Character.AI introduced Prompt Poet to make writing complex prompts for LLMs easier. Instead of messing around with basic f-strings, Prompt Poet uses YAML and Jinja2, making it more powerful and organized. This helps them handle the huge number of prompts they create every day. It also makes it easier for both developers and non-developers to work on and improve prompts.
PyTorch launched torchchat to run LLMs like Llama 3.1 smoother and faster locally on devices, including laptops and even mobile phones. Torchchat builds on previous PyTorch work and offers key features like export, quantization, and evaluation tools, making it easier for developers to build local LLM inference solutions. The project offers flexibility with Python, C++, and mobile device compatibility.
The AI robotics company Figure revealed their latest creation: Figure 02, a significantly upgraded humanoid robot with a completely redesigned hardware system. Figure 02 can engage in spoken conversations with humans by leveraging OpenAI’s technology. Figure 02 has 3x the computation & AI inference available on-board compared to Figure 01. This enables real-world AI tasks to be performed fully autonomously. It also houses a powerful battery pack that gives it a 50% increase in operational time compared to its predecessor.
OpenAI released Structured Outputs in its API that ensure model-generated outputs will exactly match JSON Schemas provided by developers. OpenAI has introduced a new model, gpt-4-02024-08-06, which scores a perfect 100% on complex JSON schema following evals. The model is better than its predecessor (as per the LiveBench Leaderboard) and is 50% cheaper! You can now build more robust applications using the OpenAI API, relying less on workarounds and more on predictable, structured data.
A new AI platform Payman secured $3 million in pre-seed funding. It enables building AI agents that act like mini-project managers. These AI agents can plan strategies, delegate tasks to humans, and even pay them.
For instance, you create an AI agent on Payman to promote your company on Twitter, set a budget of $200, and provide some background information on the company. The AI agent then creates a detailed task outlining the requirements for crafting engaging tweets. This task is published on Payman's marketplace. A user sees the task, accepts it, creates the tweets, submits their work, and receives the $200 payment upon approval – all facilitated through Payman.
Mistral AI is making it simpler for developers to customize and deploy LLMs with new fine-tuning capabilities on its La Plateforme and the alpha release of Agents. You can now fine-tune Mistral AI’s flagship models like Mistral Large 2 and Codestral directly on their platform without any coding.
Agents framework wraps AI models with additional context and instruction to create powerful AI agents, all by giving simple instructions and demos in natural language.
Bardeen, an AI-powered automation tool, has just raised $3 million in strategic funding from Dropbox and HubSpot. This platform helps businesses build reliable AI agents and automate repetitive tasks without any coding. Bardeen shows you exactly how it will execute an automation before it runs. This allows you to review and modify the steps to prevent any unexpected errors. You can create automations using simple language instructions or by recording your actions.
Google has significantly reduced the cost of its Gemini 1.5 Flash API, expanded language support for Gemini 1.5 Pro and Flash to over 100 languages, and made Gemini 1.5 Flash tuning available to all developers.
The cost for Gemini 1.5 Flash input tokens has been reduced by 78% to $0.075 per million tokens, and output tokens by 71% to $0.3 per million tokens.
You can now fine-tune Gemini 1.5 Flash models with your own data for improved performance. Tuning is free, and the inference cost is the same as the base model.
Which of the above AI development you are most excited about and why?
Tell us in the comments below ⬇️
That’s all for today 👋
Stay tuned for another week of innovation and discovery as AI continues to evolve at a staggering pace. Don’t miss out on the developments – join us next week for more insights into the AI revolution!
Click on the subscribe button and be part of the future, today!
📣 Spread the Word: Think your friends and colleagues should be in the know? Share Unwind AI and let them join this exciting adventure into the world of AI. Sharing knowledge is the first step towards innovation!
🔗 Stay Connected: Follow us for updates, sneak peeks, and more. Your journey into the future of AI starts here!
Reply