Mastra

We wrote the book on building AI Agents... literally. Well not we, but my cofounder Sam Bhagwat did. Learn why he did it and what Mastra learned during the w25 YC batch.

Was blown away reading this incredibly detailed deep dive by Pragyan Tripathi into building agentic workflows with Mastra!

and now Mastra has more Github stars than LangGraph

Looking forward to checking out Mastra’s AI Agent Cloud Platform! If you're into building AI agents, check out their open-source framework Mastra - hoping to get access to their Private Cloud Beta and start building agents with typescript

Building an AI Agent with Reasoning Capability is as simple as ?? We are excited to share this demo built by Shane Thomas using Mastra and Fireworks AI for DeepSeek R1 model. Mastra is a TypeScript Agentic AI Framework that lets you build intelligent agents with persistent memory, robust state management, contextual data integration, and transparent tracking. Checkout the project here: https://lnkd.in/dvfk-yqe

The saga continues ?? Almost 20 years ago, Sam Bhagwat and I had these pictures taken. Aside from the bad haircuts (actually Sam, your’s might have been cool), we were just young college students that joined Phi Kappa Psi Fraternity. Fast forward. I backed Mastra, many other fraternity brothers also did, my venture fund followed suit, then a bunch of other investors from my network also did. Now they have one of the fastest customer growth rates I’ve seen in software ??, and will be one of the top companies (dare I say top) at Y Combinator demo day. Surround yourself with great people who inspire you and make you excel and great things will happen. I’ve made a ton of mistakes in life, but that is the one thing I’ve always gotten right. As an added benefit, it’s been great jamming with Sam on fundraising strategies and it feels like we’re back in college again. Congrats on all the early success brother!

Mastra is the open-source JavaScript framework for building agents. Companies are using Mastra to automate support, build CAD diagrams, scrape the web, do medical transcription, automate medical transcriptions, generate financial documents, and a wide variety of other business tasks. As large language models like ChatGPT and Claude get better, software companies are increasingly building chat-based assistants into products, as well as full-fledged agents: software that does digital work. But building these products, and getting them to perform at a high-enough level of quality to go live, is difficult. Mastra makes it a lot easier. Mastra makes it possible to give agents tools (functions they can execute), long-term and short-term memory of user chat, and construct knowledge bases for agents to reference using RAG. It also helps developers ensure their agents are doing what's expected with tracing, evals (AI-powered tests), and a local development environment. The team has spent the last decade in the open-source JavaScript world as the co-founder and key engineers of Gatsby, the open-source React web framework. Their goal is to make it as easy to build and ship an agent, as it is to build and ship a website. Congrats on the launch, Sam Bhagwat, Shane Thomas, and Abhi Aiyer! ?? https://lnkd.in/deG8Y8kT

We are giving agents a voice ?? ?? You might be reading this tweet but humans don’t just communicate via text - we talk, too! Introducing Mastra voice: give your agents the ability to have real-time voice conversations with users. Mastra voice supports multiple speech providers so you can not only pick the best one for your app but mix-and-match multiple providers. Check out the demo video Yujohn Nattrass made.

The difference between AI success and failure often isn't the model—it's the context you provide. - superglue (YC W25)'s Stefan Faistenauer reveals how their data transformation tool handles ambiguous scenarios - When AI faces multiple valid options (like which of 3 prices to display), it needs additional context - Real-world case: GradeWiz dramatically improved grading by simply telling AI which university it was grading for - Why transparency about uncertainty builds user trust when AI makes difficult decisions - The simple rule: if a human would need specific information to decide, your AI absolutely requires it too #aiengineering #artificialintelligence #aiagents