Julep: A platform to manage memories, knowledge and tools for LLM apps
Hi all! We've built our fair share of LLM apps, everything from Shopify agents to real-time clones of "Samantha" from "Her". We started facing a lot of annoyances & issues repetitively with building functional apps. So we sifted through 97K posts from the OpenAI Community to confirm and find people with similar problems (& we did). It turns out that there are a handful of low-level problems that everybody who is building an AI app that works well needs to solve: statefulness to manage conversations interaction interface between multiple agents and multiple users. robust document search…
In plain words
Julep is a platform for building and managing LLM applications that addresses core infrastructure challenges. It provides tools for conversation state management, multi-agent and multi-user interactions, document search, dynamic tool calling, and model switching while preserving application state. The platform is designed for developers building production-ready AI applications who need reliable solutions to common technical problems encountered across LLM app development.
written from the facts on this page · September 2026
From the sources
In the maker’s words, at launch
Hi all! We've built our fair share of LLM apps, everything from Shopify agents to real-time clones of "Samantha" from "Her". We started facing a lot of annoyances & issues repetitively with building functional apps. So we sifted through 97K posts from the OpenAI Community to confirm and find people with similar problems (& we did). It turns out that there are a handful of low-level problems that everybody who is building an AI app that works well needs to solve: statefulness to manage conversations interaction interface between multiple agents and multiple users. robust document search dynamic tool calling from different apps swapping out different LLMs/models while preserving state, And that's exactly what we ended up making Julep to be. We built this as a prototype to solve our problems, and now we want to lower the barrier for someone to create functional and production-ready LLM apps. For more advanced stuff; We're also working on a task specification that defines a task execution framework for LLMs using a YAML format. It includes task triggers, available tools, debugging options, input and output schemas, and workflows. The main workflow is the entry point, with conditional transitions, error handling, and mapping operations for parallel processing. We want to make a Supabase-like backend for AI apps with a great DX. So if you're interested in this stuff or want a platform that you can make design decisions for, specifically for yourself, please submit PRs, become a maintainer, or join the Discord to check out cool examples we're building!
More ai this month
the category →
I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.
AI · 17d ago · simedw.com
Astute▲585Automate your B2B brand going viral, with new media creators
AI · 18d ago · company-app.joinastute.com
Hey HN, Henry from Cactus here! We previously released Cactus Needle, a 14MB agentic LLM for tool call, device use, and structured extraction for phones, wearables, smart homes, small robots and microcontrollers. We got really great feedback here, and have now incorporated the suggestions to release Needle 2. The whole model is a single 14MB binary that runs a full session in 28MB of RAM; 45m parameters at 2bit compression. Needle hits 500 tokens/sec decode speed on a Raspberry Pi 5, sits between 400-1,500 tokens/sec on VR devices like Meta Quest 3S and Apple Vision Pro, and ranges…
AI · 26d ago · cactuscompute.com
