Skip to main content

helix

Private GenAI Stack

We make Open Source models work for you. Build AI Apps with RAG, APIs and Fine-tuning. Deploy locally, in-VPC or use Helix Cloud

Try Helix Cloud Arrow right

Ship faster with a self-hosted AI App stack for everyone

Interface layer

How users interact with your AI App

  • Helix UI
  • Embed
  • SDKs
  • APIs
  • Voice

Your Helix Apps

Apps are like "GPTs" but run locally and can be version controlled

  • Prompts
  • API tools
  • GPTScript
  • RAG
  • Fine–tuning

External LLMs

Optional: GPT-4, Gemini, Claude. If you've already selected an LLM, use it

Local models

Run Llama, Mistral etc on the Helix GPU scheduler on your own hardware/cloud

Private data

Documents, web pages, APIs get connected via RAG, APIs or Fine-tuning

Easiest DevOps for Private GenAI

Production–ready GenAI for Platform Teams On K8s/OpenShift, in your VPC or simply Docker on an NVIDIA GPU

Helix API Tools

Connect and drive internal & external APIs with natural language, such as customer data and product catalogs

Knowledge

Ingest and scrape URLs, upload PDFs, word docs etc. Refresh knowledge on a regular basis

Curated Models

The best open LLMs, secure, scalable & private. Optionally pass through to closed models

Build AI Apps

Write application logic in natural language and deploy Javascript frontends for AI agents

Fine-tune models

Build your own smaller models, specialized in your domain and tooling to reduce energy and cost at scale

Helix Assistants

Embed AI assistants into your websites / apps to add a chatbot specialized with tools and knowledge

Secure to operate

Deploy privately on Kubernetes or Docker and in any cloud. Attach hybrid runners flexibly

A generated image of a swirling nebula

Take a look

Anyone can build AI Apps with the Helix App Editor. Then Evals + DevOps gets it to production

Deployment options

Helix runs anywhere, from your desktop to your cloud, or ours

Cloud

Hosted SaaS

  • Access GPT-4 level models
  • Unlimited RAG and Knowledge
  • Image Generation
  • Hosted GPTScript Apps
  • 10 fine-tunes per month
  • Set up 5 Helix Apps
  • Access all features above
  • $20/user/month premium tier
Get started
Private

On Prem / In VPC

  • Complete Data Security
  • Free < $10M annual revenue
  • Docker Desktop License
  • Source code available
  • $1K/month for up to 10 users
  • $300/month per GPU, plus support
  • Use own GPUs or external LLMs
  • Let us craft you a pilot

Try for yourself

curl -sL -O https://get.helix.ml/install.sh && bash install.sh

Installer will prompt you before making changes to your system

GenAI Experiment Pilot

A rapid 8 week program to evaluate, prototype and prove out applications of Generative AI in your business. Establish viability of OSS AI for your private use case.

Price for 8 week program: $50,000.
Email Chris to get started.

Week 1–2
  • Workshop business use cases
  • Select 2–3 PoCs
  • Deploy Helix in your environment
Week 3–4
  • Identify APIs, tools & integrations
  • Or data for RAG & fine-tuning
  • Kickoff PoC builds in collaboration with your dev team
Week 5–6
  • Hackathon to build PoCs
  • Define evaluation criteria
  • Ship PoCs to stakeholders
Week 7–8
  • Present PoCs to your executive team
  • Identify most promising ones
  • Define road to production
Head and shoulders photograph of Romain Klein Rescale

‘I've been hooked on the fact you can deploy Helix on your private infrastructure’

Romain Klein, Rescale
Our team

Gen AI experts with over three decades of industry experience at the coalface of start-ups, consulting and data AI.

Headshot of Phil Winder
Phil Winder

Full Stack AI Engineer

Headshot of Kai Davenport
Kai Davenport

Full Stack AI Engineer

Headshot of Obi Obianuo
Obi Obianuo

Full Stack AI Engineer

Headshot of Chris Sterry
Chris Sterry

COO

Headshot of Luke Marsden
Luke Marsden

CEO

Headshot of Matt Barker
Matt Barker

Advisor

Trusted by
Bayes Price Microsoft Google AWA Alberta University Sky Canonical Pinterest Docker

Not convinced?

We’re always happy to help, but if you’ve got a burning question you need answering, try one of the following channels

FAQ Discord Newsletter Try it!