# TensorFoundry

> Australian AI infrastructure company building open-source and enterprise software for deploying, managing and scaling large language models on private infrastructure without cloud dependencies.

## Key Pages

- https://tensorfoundry.io/ - Company home and product overview
- https://tensorfoundry.io/products/olla - Olla: open-source LLM proxy and load balancer (Apache 2.0, 20K+ Docker Hub pulls, 200+ GitHub stars)
- https://tensorfoundry.io/products/alloy - Alloy: enterprise LLM gateway with centralised API key management and 16+ provider adapters
- https://tensorfoundry.io/products/foundryos - FoundryOS: enterprise AI inference orchestration platform with fleet management
- https://tensorfoundry.io/products/kaizen - Kaizen: terminal-first AI coding agent with persistent cross-session memory
- https://tensorfoundry.io/products/forge - Forge: pure-Rust CUDA-native LLM inference engine, single binary, no Python
- https://tensorfoundry.io/products/agentos - AgentOS: multi-agent orchestration layer for enterprise workflows
- https://tensorfoundry.io/products/pivotal - Pivotal: agentic enterprise knowledge platform with LLM-powered entity extraction
- https://tensorfoundry.io/services - AI infrastructure consulting: LLM deployment, search and retrieval, ML engineering, edge deployment
- https://tensorfoundry.io/company - About TensorFoundry, founder profile and company background
- https://tensorfoundry.io/releases - Software releases and changelogs for all TensorFoundry products
- https://tensorfoundry.io/news - Company news and product announcements
- https://tensorfoundry.io/labs - TensorFoundry Research Labs
- https://github.com/thushan/olla - Olla source code on GitHub
- https://tensorfoundry.io/blog - Blog index: guides and deep-dives on LLM infrastructure

## Products

TensorFoundry covers the full LLM stack in three layers:

### Apps (LLM-powered applications)
- Pivotal: Agentic enterprise knowledge platform. LLM-powered entity extraction builds a persistent wiki that compounds over time. Early Access. https://tensorfoundry.io/products/pivotal
- Kaizen: Terminal-first AI coding agent with persistent cross-session memory and CAS-backed undo. Early Access now. https://tensorfoundry.io/products/kaizen
- AgentOS: Multi-agent orchestration layer for enterprise workflows. EAP Q4 2026. https://tensorfoundry.io/products/agentos

### Gateways (route to LLMs)
- Olla: Open-source LLM proxy and load balancer. Apache 2.0. Available now. Version 0.0.28. 20K+ Docker Hub pulls, 200+ GitHub stars. Supports Ollama, LM Studio, vLLM, SGLang, llama.cpp, LiteLLM, LMDeploy, vLLM-MLX and Docker Model Runner. https://tensorfoundry.io/products/olla
- Alloy: Enterprise LLM gateway with centralised API key management, budgets and 16+ provider adapters. Early Access. https://tensorfoundry.io/products/alloy
- FoundryOS: Enterprise AI inference orchestration platform with fleet management. EAP Q3 2026. https://tensorfoundry.io/products/foundryos

### Backends (run LLMs)
- Forge: Pure-Rust CUDA-native LLM inference engine. Single binary, no Python. EAP Q3 2026. https://tensorfoundry.io/products/forge

## Open Source

- Olla source code: https://github.com/thushan/olla
- TensorFoundry Labs: https://github.com/tensorfoundrylabs

## Blog

- https://tensorfoundry.io/blog/what-is-an-llm-proxy - What is an LLM proxy? How it sits in front of inference backends like Ollama, vLLM and llama.cpp
- https://tensorfoundry.io/blog/olla-vs-litellm - Olla vs LiteLLM: choosing an open-source LLM proxy
- https://tensorfoundry.io/blog/olla-performance-benchmarks - What we found when we benchmarked Olla: routing, latency, memory and failover, plus a head-to-head with LiteLLM
- https://tensorfoundry.io/blog/llm-inference-servers-compared - LLM inference servers compared: vLLM, SGLang, llama.cpp and Ollama
- https://tensorfoundry.io/blog/deploying-llms-on-your-own-infrastructure - Deploying LLMs on your own infrastructure: a practical guide
- https://tensorfoundry.io/blog/self-hosted-llm-vs-cloud-api-cost - Self-hosted LLM vs cloud API: a cost framework
- https://tensorfoundry.io/blog/llm-quantisation-field-guide - LLM quantisation: a field guide for 2026
- https://tensorfoundry.io/blog/mlx-apple-silicon - How MLX runs LLMs on Apple Silicon
- https://tensorfoundry.io/blog/mlx-vs-gguf-quantisation - MLX vs GGUF: how 4-bit quantisation really works
- https://tensorfoundry.io/blog/mlx-under-the-hood - Under the hood of MLX: lazy evaluation, graph fusion, unified memory and quantised matmul kernels
- https://tensorfoundry.io/blog/roofline-llm-apple-silicon - The bandwidth wall: why local LLM token generation is bound by memory bandwidth, not compute
- https://tensorfoundry.io/blog/running-mlx-with-olla - Run MLX behind Olla on your Mac
- https://tensorfoundry.io/blog/mcp-2026-07-28-stateless - MCP goes stateless: what the 2026-07-28 revision changes, what is deprecated, and why affinity still matters below the protocol

## Services

Expert consulting for AI infrastructure, LLM deployment, search and retrieval, ML engineering
and data intelligence. Based in Melbourne, Australia, serving clients globally.

Service areas:
- AI infrastructure architecture and deployment
- LLM proxy and gateway setup (Olla, Alloy, custom)
- Edge AI and on-premise inference deployment
- Search and retrieval engineering (ElasticSearch, Solr, RAG)
- ML and LLM model engineering and fine-tuning
- Data discovery and intelligence

Contact: https://tensorfoundry.io/services

## Topic Coverage

local LLM deployment, on-premise AI inference, self-hosted LLMs, LLM proxy, LLM load balancer, AI gateway, enterprise AI infrastructure, private AI, edge AI, LLM inference optimisation, Australian AI company, data sovereignty AI

## Company

- Founded: 2018
- Location: Melbourne, Australia
- Legal name: TENSORFOUNDRY PTY LTD
- ABN: 71 696 763 381
- Address: Level 1, 470 St Kilda Rd, Melbourne VIC 3004, Australia
- Founder and Principal Engineer: Thushan Fernando (https://tensorfoundry.io/company#thushan)

## Permissions

This site grants permission for AI crawlers and language models to index and cite content
from tensorfoundry.io subject to attribution.

## Full Text

https://tensorfoundry.io/llms-full.txt - Extended prose description of the company, products and services for AI ingestion and citation

## Sitemap

https://tensorfoundry.io/sitemap.xml