Modular AI infrastructure, built for your operation
The company turns on the modules it needs, uses internal credits as it goes and works with Lumeis engineers inside the operation.
Five modules, one infrastructure
Each module works on its own and talks to the others. The first process uses only what it needs.
Integration
Connects ERP, CRM, data warehouse, HR and finance systems and spreadsheets into one layer that refreshes on its own.
Context
Knowledge base and ontology: the rules, definitions and relations of the business in one place.
Agents
They answer, alert and investigate in Teams, WhatsApp and on the web, citing the source.
Apps
Cockpits and tools from Lumeis Software, built for the way each department works.
Governance
Permissions, audit and usage limits by department, across every module.
One unit of usage for the whole company
Everything that runs on Lumeis OS uses internal credits. The same unit covers every module, AI model and department, so usage can be compared, capped and reallocated.
How much each action uses, in proportion
- One question to the contextSomeone on the team asks and the agent answers with the source.
- One daily alert routineEvery morning, the agent reads the sources and flags only what changed.
- One root-cause investigationThe agent crosses several sources to explain a number that is off.
Limits by department
Each department has a monthly cap. Near it, the owner gets a warning.
Usage in plain sight
A dashboard shows usage by department and agent, every day.
Fewer credits with better context
The better the context, the less the agent reads to answer.
The same Lumeis OS, wherever your IT needs it
Modules, credits and governance work the same in all three models.
- Cloud
Lumeis cloud
An environment managed by Lumeis, with continuous updates and each customer isolated.
- BYOC
In your cloud
Runs in your company’s cloud account. Data stays in your environment and IT controls access.
- On-premises
On your servers
Installed in the company data center, maintained by the Lumeis team.
Lumeis engineers inside the operation
Lumeis FDEs, forward deployed engineers, work with your team from the first data source to the process in production.
They map the process
They learn how the department really works and agree on the metric the process has to move.They build and run it
They connect the systems, build the context and put agents and apps into production.They hand over the know-how
They train your team to build on Lumeis OS and keep the context up to date.
Value measured in every process
Every process starts with an agreed metric and is measured against the before. Context is what makes the numbers work.
- 4.5×
more right answers when the model works with an ontology
- −98.7%
tokens per task when the agent reads only what it needs
- 1 day
a week given back to each person who was searching for information
See the studies behind these numbers
Lumeis Bench gathers public research on accuracy, token usage and time, and shows how Lumeis OS gets to those gains.
Frequently asked questions
Do I need every module?
No. The first process uses only the modules it needs. The others come in when a new process calls for them.
Can I cap usage by department?
Yes, and everything is configurable. You can set caps by department, user, agent or app, choose when the owner gets a warning, require approval above the limit and reallocate credits between departments during the month.
Which AI models does Lumeis OS use?
Models from several providers, at the same time. Each task can use the best-suited model, or follow the rule your IT sets by department, agent or data type.
Can I use the model subscriptions the company already has?
Yes. Lumeis OS manages subscriptions from several providers in one place. You can bring the company keys, combine them with Lumeis ones and switch models without rebuilding agents and apps.
How do I track usage?
In a dashboard that shows usage by department, user, agent, app and model, every day. The data can be exported to the company BI.
Do credits cover every module?
Yes. Integration, context, agents and apps use the same unit, so the company compares usage across departments and processes on the same scale.
Can I change the deployment model later?
Yes. Modules and context are the same in the Lumeis cloud, your cloud and your servers, so you can start in one model and move to another.
How long until the first process is in production?
Four weeks, with Lumeis FDEs inside the operation and the result measured against the metric agreed at the start.
Is company data used to train models?
Never. Neither by Lumeis nor by the model providers. Permissions and audit apply to every access.
Who owns what gets built?
Your company does. Agents, apps, context and data stay with you and can be exported at any time.
Design the first process with us
In 30 minutes, we map the process and the modules it needs.