Robots making robots

PLUS: ChatGPT connectors for G-Drive and Slack

Author

Barsee
18 Mar

Together with

Howdy again. It’s Barsee, and welcome back to AI Valley.

Another day, another AI adventure.

Today’s climb through the Valley reveals:

  • Mistral's latest compact AI beats Gemma 3 and GPT-4o Mini

  • Robots making robots

  • ChatGPT connectors for G-Drive and Slack

  • Plus trending AI tools, posts, and resources

Let’s dive into the Valley of AI…

PEAK OF THE DAY

Mistral's latest compact AI beats Gemma 3 and GPT-4o Mini

French AI startup Mistral AI just launched Mistral Small 3.1, a compact yet powerful model that outperforms Google's Gemma 3 and OpenAI's GPT-4o mini while running efficiently on consumer hardware.

Image Source: Mistral AI

Here’s what you need to know:

  • Mistral 3.1 boasts improved text generation, enhanced multimodal understanding, and an expanded context window of up to 128,000 tokens. This allows it to handle complex tasks that require retaining context over long conversations, detailed document analysis, or multi-step reasoning.

  • With only 24 billion parameters, the model is lightweight enough to run on a single GPU or devices with 32GB RAM, making it accessible for developers without requiring extensive hardware.

  • Operating at a speed of 150 tokens per second, Mistral 3.1 is ideal for applications requiring fast responses, such as virtual assistants, real-time customer support, and interactive chatbots

  • The model demonstrates state-of-the-art performance across multiple benchmarks, surpassing Gemma 3 and GPT-4o mini, particularly in MMLU (general knowledge), HumanEval (coding), and GPQA (question answering), despite having fewer parameters.

  • Mistral 3.1 is released under the Apache 2.0 license, allowing developers to freely use, modify, and build upon it. Both base and instruct checkpoints are also provided, enabling fine-tuning for specialized fields like healthcare, legal analysis, and customer service.

  • The model is accessible via Hugging Face and Mistral AI's developer platform, La Plateforme. It will soon be integrated into cloud services like Google Cloud and Microsoft Azure AI Foundry.

Why it matters: 

Mistral Small 3.1 represents a significant milestone in AI development by offering superior capabilities while remaining lightweight and open-source. Its versatility across applications ensures that it will play a crucial role in shaping the future of generative AI solutions across industries.

3x faster than typing using AI

Image Source: Flow

Meet your new productivity hack: Work smarter, not harder with an app that transforms jumbled thoughts into clear communication. Flow captures your words in real time and formats them so you don't have to. Flow automatically edits out filler words and clutter. Want a different tone or a quick AI-driven rewrite? Just ask Flow.

  • 3x Faster Than Typing – Get more done in less time.

  • Natural Voice – Flow sounds like you, not a stilted bot.

  • Auto-Edits & Formatting – No more typos or punctuation errors.

  • 100+ Languages – Communicate globally without missing a beat.

Because typing is so last decade. Stop typing, start speaking—Flow will handle the rest.

*This is sponsored

VALLEY VIEW

California-based Figure is building a high-volume BotQ facility to produce 12,000 humanoid robots annually, scaling with vertical integration. Advanced systems like MES, PLM, and ERP will streamline operations, optimize inventory, and boost automation. Transitioning from Figure 02 to the production-ready Figure 03, the company uses faster methods like injection molding and die-casting. A hybrid workforce of robots and humans, plus AI-driven automation, ensures efficient mass production.

OpenAI has initiated beta testing for ChatGPT Connectors, a feature that enables ChatGPT Team users to integrate applications like Google Drive and Slack directly into ChatGPT. This integration allows the AI to access and utilize internal documents, spreadsheets, presentations, and Slack conversations to provide more informed responses.

Nvidia's GTC 2025 is happening from March 17–21 at San Jose's McEnery Convention Center, featuring 1,000+ sessions, 2,000 speakers, and 400 exhibitors. Highlights include the Blackwell Ultra GB300 series reveal, boasting 50% more memory and performance. A dedicated quantum computing day explores practical applications. CEO Jensen Huang’s keynote is set for Tuesday at 10 a.m. PT, livestreamed on Nvidia's website and YouTube.

Cursor has introduced Claude Max, its most advanced AI model designed for developers managing large projects. Built on Claude 3.7, it offers a 200k context window and supports up to 200 simultaneous tool calls, allowing extensive code edits and full codebase analysis in one go. Claude Max follows a pay-as-you-go pricing model at $0.05 per prompt and tool call, making it ideal for professional developers rather than casual users.

Zoom has added AI agents to its AI Companion, enabling it to handle multi-step tasks like managing calendars, summarizing meetings, and generating video clips. New features include Zoom Tasks, which identifies and completes action items from meetings, chats, and emails, and a voice recorder that transcribes and summarizes in-person conversations. These updates will roll out gradually through mid-2025.

Roblox is launching and open-sourcing Cube 3D, a generative model that creates detailed 3D objects or meshes from simple text prompts. The tool allows users to customize, build plug-ins, and train the model with their own data. Roblox also announced upcoming AI tools for text generation, text-to-speech, and speech-to-text, set to release in the coming months.

Nvidia is expected to unveil details about its latest AI chip, Rubin, during its annual developer conference today. Rubin will target large data centers supporting AI training and is anticipated to enter production this year, with broader availability next year. CEO Jensen Huang may also address updates on the delayed Blackwell chip and Nvidia's broader AI strategies.

TRENDING TOOLS

  • Gemini in Android Studio > Code from images, debug with screenshots.

  • NeuraVid > Find & analyze videos instantly with AI-powered search.

  • Proxy DeepWork > Automate complex workflows with AI agents.

  • Interlify > Connect LLMs with your company’s own data and services without needing to write complicated code.

  • BrowserAgent > Create and run AI workflows directly in your browser.

THINK PIECES / BRAIN BOOST

VALLEY GEMS

1/

2/

3/

SUNSET IN THE VALLEY

Thank you for reading today’s edition. That’s all for today’s issue.

💡 Help me get better and suggest new ideas at [email protected] or @heyBarsee

👍️ New reader? Subscribe here

Thanks for being here.

REACH 100K+ READERS

Acquire new customers and drive revenue by partnering with us

Sponsor AI Valley and reach over 100,000+ entrepreneurs, founders, software engineers, investors, etc.

If you’re interested in sponsoring us, email [email protected] with the subject “AI Valley Ads”.