home →
Download Free
v2.0 — Coming Soon

The AI layer that actually remembers you.

UPtrim is a local reverse proxy that lives between your chat app and your LLM — adding persistent memory, smart context trimming, per-user isolation, and (in v2.0) hybrid cloud routing, autonomous subagents, and a real-time view of the whole pipeline.

🧠

Memory That Persists

NLP extracts facts from your conversations and stores them in a local SQLite index. Close the chat — it still knows who you are, what you're working on, what you prefer.

💡

One Proxy, Every Model

OpenAI-compatible front end means it drops into Open WebUI, SillyTavern, or anything else. v2.0 adds OAuth-based routing to Claude and GPT alongside your local models.

🔒

Local-First by Design

Your memory, files, history, and identity stay on your machine. Even when using hybrid mode, we send the minimum to cloud providers — and store nothing with them.

Why UPtrim exists

Most AI chat apps have a blind spot: every conversation starts from zero. You tell the model your name, your preferences, what you're working on — and an hour later, when you open a fresh chat, it's a stranger again.

At the same time, if you run a shared local LLM for your team, everyone's context crashes into one pool. Sarah asks about her meeting notes and gets Mike's Python script. That's not "smart."

UPtrim fixes both. It's a transparent middleware layer that observes every conversation, extracts what matters, stores it per-user, and injects the right memories into the right contexts when they're needed — all on your hardware, all yours.

v1.0 is live on GitHub today with a free developer key. v2.0 — building now — turns it into a full cognition layer with hybrid cloud routing, autonomous subagents, semantic retrieval, image generation, and a real-time dashboard that shows the whole pipeline.

The Roadmap

Where UPtrim came from, where it's going.

v1.0 — FoundationLive Now

Persistent memory via NLP fact extraction, context trimming, per-user isolation, file uploads, web dashboard, multi-backend support. Free developer key on GitHub.

v2.0 — Intelligence StackIn Progress

Hybrid cloud+local routing with OAuth (Claude + GPT), semantic memory & document RAG, autonomous subagents with shared scratchpad, prompt enhancement, conversation history search, image generation, and a real-time brain dashboard. $15/mo Pro tier.

v2.0 Premium — Power UserSame Release

Memory Gardener (auto dedup + contradiction resolve), Knowledge Graph, Persona Engine, TrimScript plugin engine, Claude Code subprocess offload, 1 GB uploads. $30/mo Premium tier.

?
v1.2+ — BeyondPlanning

Visual knowledge graph explorer, conversation branching & diff viewer, Ghost Agent mailbox admin, web verification tasks, team-scoped orchestration. Ideas open on GitHub.

Start Free. Scale When You're Ready.

v1.0 is on GitHub today. Every v2.0 feature ships to early paid subscribers as it drops.

Download v1.0 Free See Monthly Plans →