Mem0
A memory layer for AI agents: extracts and recalls facts across sessions so the LLM stops re-sending full history. Apache-2.0, self-hostable on Postgres and Qdrant; SSO and audit logs are Enterprise-gated.
Generative AI
Category: Generative AI under Artificial Intelligence
Chroma
AI-native embedding database for RAG: runs embedded via pip or as a Docker server, with portable SQLite/Parquet storage for clean data exit. Apache-2.0, but the OSS core ships without built-in authentication.
Voicebox
Local-first AI voice studio powered by Qwen3-TTS and six more engines. Clones voices and synthesizes speech fully on your machine, with no per-character metering, though your own GPU does the work.
Qdrant
Apache-2.0 vector database in Rust for RAG and semantic search, fully self-hostable with no copyleft strings. Multi-AZ high availability, SSO, and audit logs are gated to the managed Qdrant Cloud.
Paperclip
MIT-licensed control plane for fleets of AI agents: org charts, per-agent budgets, immutable audit logs, and approval workflows. The governance layer proprietary platforms paywall, here free and self-hosted.
Agent Zero
Self-hosted AI agent framework for autonomous code generation, command execution, and tool creation. Runs entirely within your infrastructure, with no external API dependency.
SurfSense
Private AI search connecting local LLMs to Slack, Notion, and Jira for cited sovereign search. No compliance certifications; self-hosting mandatory for regulated workloads.
Unsloth
Apache 2.0 LLM fine-tuning framework from Unsloth AI: 2x faster training on 60% less VRAM versus standard implementations; multi-GPU and full-parameter training require Pro/Enterprise tiers.
Jan
An open-source, local-first AI orchestration desktop that allows you to run LLMs entirely on your own hardware. A private alternative to ChatGPT.
OpenClaw
Autonomous AI agent with deep OS access and zero telemetry, and a supply chain risk profile that demands VM sandboxing before any enterprise deployment.
Accomplish
A local-first AI desktop agent for executing browser workflows and file management. A private, MIT-licensed alternative to cloud-bound AI agents.
Helicone
Enterprise LLM observability and gateway. A SOC 2-certified alternative to LangSmith for tracking prompts, costs, and performance with data sovereignty.
Parlant
Self-hostable AI agent control framework with no enterprise tax. VC-backed by a single corporate contributor; monitor for open-core pivot.
Open Notebook
Local AI research assistant with no cloud data transit and full document sovereignty, without NotebookLM's mandatory data exposure.
LibreChat
Self-hosted AI interface that aggregates multiple LLM providers into a single enterprise-controlled dashboard: SAML, RBAC, and full data sovereignty included.