Local & self-hosted
Run the application on your own servers, an on-site system or infrastructure managed by your IT partner. Compatible AI models can run locally when hardware capacity, licensing and the workload allow.
Local · Cloud · SaaS · PaaS
Your platform. Your choice of environment.
Run Mandem on your infrastructure or let us manage the environment. Choose the hosting, service model and AI access that fit your business, your team and your data.
Choose where it runs.
We scope the deployment around your existing infrastructure, security requirements and the people responsible for operating it.
Private cloud
A dedicated company environment with agreed access, network boundaries and operating responsibilities. Choose an appropriate hosting location and decide separately where AI processing takes place.
Public cloud
Use agreed cloud infrastructure with defined regions, access controls and maintenance. Configure the services and capacity around your workload as the business grows.
Choose how you use it.
Hosting is one decision. How your team buys, develops and operates the software is another.
Managed SaaS
Use your website, CMS and agreed assistant workflows as a managed service. Mandem handles the defined hosting, updates and support so your team can focus on using the tools.
A PaaS foundation
Build on a company platform with environments, integrations and deployment support shaped around your team. Agree which parts Mandem operates and what your developers or IT partner own.
Powerful AI, selected for the work.
We can configure access to leading hosted AI models alongside compatible open-weight models on approved infrastructure. Choose the capabilities your people need for writing, reasoning, coding and working with company information. Provider integrations, model availability, usage and processing terms are agreed as part of your setup.
Leading hosted models
Connect to capable model services through approved provider APIs. We agree the provider, account, permitted data and usage budget for your workflow.
Locally hosted models
Use compatible open-weight models on company-controlled infrastructure when local processing is the right fit. Test the model against your actual tasks and available compute.
A considered combination
Keep selected workloads local and use approved hosted models for others. Define which information may go to each provider before connecting the workflow.
Your data takes the route you agree.
Application hosting and AI processing are separate choices. A locally hosted website can still send a prompt to an external provider. A fully local workflow requires local processing, storage and supporting services throughout the agreed data path.
A setup shaped around your requirements.
These are deployment and service options we scope with you. Each project confirms the supported integrations, model access, responsibilities and acceptance checks before launch. The demonstration workspace is a starting point; additional providers, private inference and platform extensions are configured for the engagement.