Private deployments
Agents That Stay on Your Infrastructure
Run Operator on servers you own: GPUs and servers bought through Exos Private Cloud, or servers you already run. The agents, the models and the data stay on your infrastructure, and we help you host it.
Designed to Run on Exos Private Cloud
Exos, Makina's infrastructure partner, hosts and manages private clouds: GPUs and servers that belong to you, run for you alone.
Visit ExosWhy Teams Run Operator Privately
The data stays put
Documents, records and everything the agents see stay on your infrastructure. None of it passes through Makina's cloud.
Your own models
Open-weight models run on your own GPUs, chosen and evaluated for the work you hand over.
A computer for every agent
Each agent works on a computer of its own inside your network, with a browser, a terminal and files.
Approvals stay with your people
Anything with consequences waits for a person on your team, by the rules you set.
A record you keep
The journal of every step and the signed receipts are written and kept on your servers.
Workflows built for you
We build the workflows your teams hand over, with a person's approval wherever you want one.
Two Ways to Host It
Buy what the work needs through Exos, or put Operator on servers you already run. Either way the infrastructure is yours.
On Exos Private Cloud
Buy GPUs and servers through Exos. Exos hosts and manages them privately, for you alone, and we install Operator on them.
Visit ExosOn Servers You Already Run
Run Operator as a local harness on your own servers, in your data centre or your cloud account. We help you install it and keep it running.
Talk to usFrom First Call to Running on Your Servers
A private deployment is built with you, one step at a time. You decide how far we take it, and when your team takes over.
Plan a deploymentAssess
We map the work you want to hand over, the data it touches and the risks, and decide what runs where.
Size and source
We size the GPUs and servers the work needs. You buy them through Exos, or we use servers you already run.
Install
Operator, the agents' computers and the models go onto your infrastructure, inside your network's rules.
Choose the models
We evaluate open-weight models on your own tasks and keep the ones that do the work well.
Build the workflows
We build the custom workflows your teams hand over, with approvals wherever a person should decide.
Run and hand over
We run it with you, then hand it to your team, and stay on to support it as it grows.
Custom Workflows
Document intake
Agents read what arrives, enter what matters into your internal systems, and stop for a person where a decision is needed.
Confidential research
Agents research across your internal sources and write the brief on your own servers, with a link for every fact.
Back office in internal tools
Agents work through queues in the web tools your teams already use, inside your network, one record at a time.
Reports from internal data
Agents pull figures from internal systems into the reports your teams send, and ask before anything goes out.
Engineering chores
Agents work in a terminal on your internal repositories: upgrades, test runs and release notes, ready for review.
We help you host it. Makina’s consulting practice designs the deployment, chooses and evaluates the models, and builds the workflows with your teams.
ConsultingSecurity You Can Check
Run Operator on your own infrastructure or in ours. Either way it asks before it acts and writes down what it does, so you can check its work instead of taking its word for it.
More about security