RT @HasuraHQ: RAG and tool calling limit what AI agents can do! They use pre-defined workflows to achieve an outcome which doesn't work wh…

Thanks to everyone who joined our #DataAgents #AI meetup last night at @HasuraHQ headquarters in #SanFrancisco. If you missed it, no worries. Hope to see you at our next one! Follow Hasura PromptQL on @LumaHQ

PromptQL Programs are Python scripts—generated by LLMs—that read & write data, and are executed within the PromptQL Playground. 🤖 Now, with the PromptQL Program API, you can trigger them via HTTP, adding AI to your business workflows seamlessly—beyond the chat UI! Check it out here ➡️

Don’t miss the inaugural #AI #DataAgents meetup – an exclusive networking and learning event. (Plus free 🍕 pizza and 🍻 drinks!) Details here 👉

RAG and tool calling limit what AI agents can do! They use pre-defined workflows to achieve an outcome which doesn't work when user queries are dynamic. The solution - give your AI: 🗺️Query planning abilities 🧠Structured memory 🎛️Programmatic runtime Join our next Community Call where we'll walk you through this! 📅 Feb 27 | ⏰ 9 AM PT | 💻 Online 🔗 Register now →

🚀 Build AI-Ready Data Architectures with MongoDB & Hasura 📅 Feb 27, 2025 | ⏰ 11 AM ET | 🖥️ Live Webinar Tackling data sprawl, and security risks while the demand for faster code releases intensifies? Learn how you can use MongoDB + Hasura to solve these challenges. In this deep dive, we'll cover: 📏 Config-driven development ⚡ Building an instant data access later 🔗 Unified data access across diverse sources 🛡️ Enterprise-grade security 🔗 Register now 👉

🚀 Launching the first edition of Data Agents—a meetup focused on achieving complete reliability when deploying AI systems on enterprise data! 📅 Feb 6 | 📍Hasura SF We’ll explore the path towards 100% accuracy and repeatability in AI systems with real-world examples & architecture insights. Plus: 🔥 A fireside chat w/ Amit Butala (AI/ML Tech leadership at Rippling) 📊 The Enterprise Data Benchmark preview - a 650 question benchmark of real world business tasks with a performance review of the top models (GPT 4o, o1, Claude Sonnet 3.5, Deepseek r1) Drinks, networking & deep data convos—don’t miss it! RSVP 👉

ICYMI - Here are all the features we added to PromptQL recently: 🚀 Quickstart PromptQL with built-in LLM keys: Get started effortlessly with pre-configured LLM keys and $10 worth of free credits, eliminating the need for an external API key setup. 👨‍👨‍👦‍👦 Public PromptQL Playground: Share projects and threads publicly, making collaboration and knowledge-sharing seamless. ✉️ Share a thread publicly: Highlight key insights by sharing specific threads publicly with a unique link. Share interesting conversations or results without requiring viewers to log in while ensuring your project details remain private. 💽 Saved and shareable prompts: Easily guide users with predefined, shareable prompts, enabling smooth onboarding and collaborative exploration. 🐍 New program API endpoint Trigger Python-based PromptQL programs via HTTP, enabling smooth integration into existing workflows More about these features in detail here ➡️

🚀BIG NEWS for @HasuraHQ Data Delivery Network! Gain full visibility into your #API operations, boost #data observability, and enhance decision-making across your projects with the Platform Dashboard! Whether managing one or many initiatives, this tool will elevate your operational efficiency and control. See the full announcement here 👇

Imagine an AI customer service assistant refunding $50k instead of $500 because of a misplaced decimal. With today's AI, this scenario is a possibility. If you expect AI assistants to take on more complex, sensitive, high-impact tasks, you're going to need to build in accountability, transparency, and safety into the workflows. Human-in-the-loop (HITL) systems can mitigate all these challenges. We dive into the different scenarios where you need HILT, how to think about them, and how you can implement them with PromptQL. Read more about it here➡️

We're previewing a lot of of PromptQL updates at Thursday's Community Call (Jan 30): ✍️ Editable query plans, a new PromptQL Programs API for repeatable execution 🧱Improved AI primitives ⏲️ Extended timeouts 🌐 Support for PromptQL on Private DDN 📨 Thread sharing for effortless collaboration Don’t miss it. Register now ➡️

RT @jamdotdev: An AI assistant that actually works w/ big data - @AnushrutGupta, lead PM @HasuraHQ

A universal data access layer is a game-changer for enterprise data strategy to enable AI. Join us tomorrow (16 Jan) for a live workshop and demo session where we’ll walk you through: 🔑 A practical data access strategy 🤖 AI enablement strategies 🛂 Data governance best practices 🔂 Data quality feedback loops ♻️ Virtuous Data cycles You'll learn practical tips, frameworks, and real-world examples to help you unify enterprise data and enable AI feature development across your team/org . Register ➡️

PromptQL got ~100% on a difficult RAG benchmark! Join us on our Community Call this month (30 Jan) to find out what makes this benchmark challenging and why we need a reliable query planning approach to perform well on this benchmark. We'll also announce 6 new PromptQL features! Register ➡️

Don't miss Anushrut Gupta's talk at @jamdotdev's AI Night at GitHub today! He'll demo PromptQL - a data access agent that achieves 5x higher accuracy and repeatability when connecting LLMs to business data. Details ⤵️

AI agents often fall short: long dev cycles, inconsistent outputs, and users lose trust from just a few bad answers. Join us LIVE on Jan 9th and see how you can use PromptQL to solve these problems. 🚀 Register now ➡️