LastMile AI

OpenAI has standardized on Anthropic's Model Context Protocol- MCP ?? For those who are just getting started with MCP and are wondering: - What is MCP? - Why is it so important? - Where can I use it? Check out our latest Backprop podcast where our co-founders Sarmad and Andrew walk you through MCP and MCP with Agents. https://lnkd.in/eJE3Tzpd

?????? ?????????? ?????? ?????????????????? ?? Calling NYC-based devs - @LastMile AI is hosting an MCP Agent (https://lnkd.in/d7-vTJWA) hackathon this ?????????????? from ??:????-??:???????? at ?????????????? ??????????????! Join us and other developers to learn from, build with and showcase your MCP agents. Bring your laptop to get involved. Food will be provided. ???Registration is limited:?https://lu.ma/uysalrod

There's so much left to build, and we are just getting started ??

OpenAI’s GPT-4o image generation is incredible! Character consistency, detailed instruction following, and inpainting are all amazing improvements to image generation. Following the trend, we tried: “Generate a Studio Ghibli version of this photo.” Here’s the LastMile AI team getting ice cream in Ghibli-fashion?? To create your own Ghibli like image, just upload a photo on your ChatGPT web or mobile app, and drop in that prompt.

?????? ?????????? ???????? ???????? Join us on ???????????????? from ??:???????? ???? ??:???????? ?????our biweekly virtual meeting to demo your MCP agents, get help building your MCP agents and meet other developers! Be sure to register ahead of time so we can get you into the Zoom! Link in comments below??

LastMile AI is Hiring! Building robust and controllable AI systems is not easy. It requires good foundation models, good application architecture, and good evaluators to continuously measure and improve performance. Once in production, it requires guardrails and monitoring. Above all, it requires staying nimble to adopt new techniques being invented in the fastest-paced period in the history of technology. We started LastMile AI to help enterprises build, test and deploy AI applications. Our users are developers just like us. Our roadmap is ambitious. We are hiring across the board for exceptional engineers who are interested in taking on this “last mile” challenge.? We are focused on: ? ? Foundational ML research – small language models for eval tasks, synthetic data generation,? ? Developer productivity – mcp-agent for building effective agents with Model Context Protocol, and? ? Tons of infrastructure work to orchestrate and deploy the backend for an AI Ops platform. Get in touch if any of this excites you, and you want to work alongside some world-class engineers and researchers! We have open roles in the comment below, but even if it isn’t a perfect match, let us know why you’re passionate about this space and we’ll chat. Let’s build something groundbreaking together. ?? #Hiring #AI #Engineering https://lnkd.in/eY4mzD-7

Agents! Agents! Agents!

Episode 6 of Backprop is Live – The Infrastructure of AI Agents with Salman Paracha With the new release of OpenAI’s new?Agentic tools, AI agents have been top of mind for a lot of teams. But while the idea of?autonomous workflows?is exciting, the reality of building and scaling AI agents is a whole different challenge. In this episode,?Sarmad sits down with Salman Paracha, CEO of?Katanemo, to break down the?infrastructure behind AI agents—what’s working, what’s not, and what needs to evolve. Salman has built and led teams at?Oracle and Amazon?before co-founding Katanemo, where he’s developing intelligent infrastructure for AI-native applications. He and Sarmad also dive into?MCP agents, sharing insights from Sarmad’s own work in this space and how teams are thinking about deploying and scaling AI-driven systems. If you’re working with AI agents—or just trying to figure out how they fit into your stack—this is one to check out. Now streaming on YouTube and Spotify.????

New Anvil - Document SDK #Build vs #Buy episode drop. Learn about the company building tools for getting your toy #AI models production ready! Andrew Hoh is the cofounder and #CPO of LastMile AI. He also happens to be a fellow Gradient Ventures founder and former PM Meta, Airbnb and Microsoft. Learn about his build vs buy decision framework and how he fell sideways into product management. ?? Full episode link ?? https://lnkd.in/g8Peh9vb

Last week, we introduced?Experiments in AutoEval, a powerful way to?track, compare, and validate improvements?in generative AI applications. If you haven’t tried it yet, now’s the time.?Watch this demo Andrew put together?and see how it works in action. Experiments can systematically improve any AI system with?highly configurable parameters—whether it’s?tuning a retrieval strategy for RAG?or?refining system instructions for agents. Keep experimenting!???