Technologies
Open-Source
Own the stack when it matters.
Open source gives you control: your data stays on your hardware, there is no per seat licence, and nothing disappears when a vendor changes direction. We design, build and run open source AI where sovereignty and cost make it the right call.
YOUR OPEN-SOURCE ESTATE · MAPPED
RUN IT YOURSELF
Linux
Docker
LOCAL MODELS
Ollama
LM Studio
OPEN AGENTS
OpenClaw
Hermes Agent
6 building blocks · 3 layers
owned and governed, not rented
Products and services
What we deliver across the open source stack.
Six building blocks, three layers. Start with the layer where your pain lives.
Run it yourself
The foundations under everything
Linux
The operating system under almost everything: servers, containers and most of the cloud.
Docker
Containers that package software so it runs the same on a laptop, a server or the cloud.
Local models
AI on your own hardware
Ollama
Runs open models on your own hardware with one command. The standard for local AI.
LM Studio
A desktop app for downloading and chatting with local models. The easy way to test what open models can do.
Open agents
Autonomy you can self host
OpenClaw
A widely used open source personal AI agent: messaging first, self hosted and endlessly extensible.
Hermes Agent
Nous Research’s open source agent with persistent memory and self built skills. It learns your environment the longer it runs.
Why CG TECH
Why CG TECH for open source AI.
Open source is free to download and expensive to get wrong. We design, build and run the stack, then train your people so it lasts.
Sovereignty by design
Your data stays on infrastructure you control, which matters in regulated and sensitive environments.
No licence lock in
No per seat fees, no per token surprises and no vendor deciding your roadmap.
Build and run, not just advise
We design the stack, harden it and run it with you until your team owns it.
Adoption that sticks
Training and change support are part of every engagement, because a platform nobody uses is a cost, not an asset.
Where open source fits.
Data sovereignty, predictable costs at steady volume, air gapped or regulated environments, and teams that want full control of their stack.
Where it does not.
Frontier capability per dollar still favours hosted models for most teams, and self hosting means owning the operations. If nobody in your organisation can run it, managed platforms win. You get one honest recommendation, not a default.
What good looks like
Real projects, real outcomes.
Questions we hear a lot
Common questions about open source AI
What does open source AI mean in practice?
It means the model and the software around it are free to download, run and change, on hardware you control. Your data stays on your own machines, there is no per seat licence, and nothing disappears when a vendor changes direction.
Is open source AI good enough for business use?
For bounded, well defined tasks, yes, and it keeps improving. For frontier reasoning, hosted models still lead. Most organisations land on a hybrid: local where sovereignty or volume demands it, hosted where capability demands it.
What does open source actually save us?
There is no per seat or per token licence. You pay for hardware and for the people who run it. At steady, high volume the sums often favour open source, and we model that honestly before you commit.
Is self hosted AI more secure?
Your data never leaves your network, which is a real advantage. But you own patching, hardening and access control. We set that up properly rather than leaving it as an afterthought.
Can you support what we already run?
Yes. We design, harden and support self hosted stacks, and we run this tooling ourselves every day.
When is open source the right call, and when is it not?
It is the right call when data cannot leave your walls, when volume makes per seat licensing painful, or when you need to keep running a model long after a vendor retires it. It is the wrong call when you want the strongest reasoning available and have nobody to look after a server.
What size organisation do you work with?
Our sweet spot is 50 to 500 seats. At that size the licence maths starts to matter, and there is usually someone in house who can own a server with our support behind them.
Do we need our own hardware?
Not always. Open models can run on your own machines, in your own cloud tenancy, or with a hosted provider. We size the options against the workload before anyone buys anything.
Ready when you are
Thinking about open source AI? Let us talk.
A discovery session maps your use cases, your risks and your quick wins. You keep the plan either way.
What to expect
- A consultant replies within 4 business hours
- Session booked to understand your requirements
- We will provide you with a fixed price quote