Private deployment · on-premise
RevOS in your cloud, on your lakehouse.
Run RevOS as a private deployment in your own cloud account or data center. It is single-tenant, and it builds on the lakehouse, warehouse, or database you already have. More than 25 backends are supported, including BigQuery, Snowflake, Databricks, Microsoft Fabric, Redshift, ClickHouse, and Postgres. Your data, your metrics, and your AI agents stay inside your environment, and so can AI inference, with your own key or a model endpoint on your premises.
30 minutes with our engineers on your architecture and your security questionnaire.
Who it's for
A governed data foundation where your data already lives
Teams in finance, healthcare, security, and the public sector want trusted metrics and AI agents on their commercial data. Most platforms that offer this pull the data into their own cloud first, which is where the security review stops. A private deployment runs RevOS next to your data and inside your perimeter, so you do not have to choose between modern and secure.
Customer and revenue data cannot sit in a vendor's cloud
Data-residency rules, customer contracts, vendor-risk policy
Your cloud, your region, single-tenant.
You already chose a lakehouse and will not run a second one
Platform standards, committed spend, one governance model
RevOS runs on the one you have.
AI agents cannot read data that leaves your network
AI policy, model-provider review, no external egress
Agents query a governed layer inside your environment.
Bring your own lakehouse
Built on the lakehouse you already run
RevOS does not ask you to adopt a new warehouse. It uses yours as the backend. Connectors land data there, and the Semantic Model, the Metric Store, scores, and segments are computed there. Everything stays in standard tables your team can query, back up, and govern with the tools it already uses.
Google BigQuery
Snowflake
Databricks
- Microsoft Fabric
Amazon Redshift
ClickHouse
PostgreSQL
- Microsoft SQL Server
MySQL
Oracle
- Amazon Athena
Trino
Presto
DuckDB
- MotherDuck
Hive / SparkSQL
SingleStore
- Apache Pinot
Apache Druid
- Firebolt
- Vertica
MongoDB
SQLite
- Materialize
- RisingWave
- QuestDB
- ksqlDB
Cloud warehouses, query engines, and the databases you host yourself are all supported. If yours is not listed, tell us which engine and we will confirm the fit when we scope the deployment.
How it deploys
From scoping call to running platform
Your platform and security teams stay in charge of the environment at every step. Our engineers work alongside them.
Scope it with your platform team.
We agree where RevOS runs (your cloud account or your data center), which lakehouse it builds on, how users sign in, and where AI inference runs: with your own key at a provider you have approved, or on a model endpoint inside your premises.
Clear it with your security team.
We complete your security questionnaire, walk your reviewers through the architecture, and document every network path the deployment uses. Most reviews get shorter at this point, because the data stays where your controls already apply.
Deploy inside your perimeter.
RevOS ships as versioned container images that run on Kubernetes. Your team installs them with our engineers on the call, behind your own network rules, secrets management, and access policies.
Connect your sources and your lakehouse.
Connectors run inside your network and land data in your lakehouse. The semantic model, metrics, scores, and segments are computed there, in tables you can query directly.
Control stays with you
What runs in your environment and what RevOS provides
Everything that touches your data runs inside your environment, behind the controls your security team already operates. RevOS ships the software, and your team runs it. The questions that stall a vendor review (where the data goes, who can reach it, how you would leave) all get the same answer: it stays with you.
01
Runs in your environment
your cloud or data center
- RevOS application and API, single-tenant
- Connectors and sync jobs
- Semantic Model and Metric Store
- Scoring, segments, and Actions
- MCP server for your AI agents
02
Yours, not held by RevOS
referenced by the deployment
- Your Kubernetes cluster
- Your lakehouse, warehouse, or database
- Storage and backups
- The application database
- Your identity provider, through single sign-on
- Your model endpoint and API keys for AI inference
- Credentials and encryption keys
03
RevOS provides
release artifacts, not the runtime
- Versioned container images and deployment manifests
- RevOS Agent, the built-in AI agent
- The revos CLI for configuration as code
- Architecture documentation for your security review
- Support under an Enterprise SLA
The questions your security team will ask, answered
- 01
Where does our data go? Nowhere new. Raw data, models, and metrics stay in the lakehouse you already run, in your cloud and your region.
- 02
Who else is on the infrastructure? Nobody. One deployment serves one organisation, with its own application, database, and compute.
- 03
What can the vendor reach? Nothing by default. The deployment needs no inbound connection from RevOS, and our engineers get access only when you grant it.
- 04
What do AI models see? Only what you allow. Inference runs with your own key at a provider you have approved, or on a model endpoint inside your premises.
- 05
Can we prove who saw what? Yes. Single sign-on, role-based access, and an audit history cover every person and every agent query.
- 06
What if we leave? Your data is already in standard tables in your own lakehouse, and your metric definitions are code in your own repository.
What's included
The full platform, not a reduced edition
A private deployment includes the same product as RevOS cloud. What changes is where it runs.
Connectors and Data Lake
- Managed connectors for CRM, billing, product, and support data
- Sync jobs that run inside your network
- Data lands in your lakehouse as standard tables
Semantic Model
- Metrics and dimensions defined as code
- Changes reviewed in pull requests and applied with the CLI
- One model for dashboards, BI tools, and agents
Metric Store
- A catalog with an owner and status for every metric
- Version history for each definition
- Lineage from source to dependents
Scoring, segments, and Actions
- Health and lead scores built on governed metrics
- Dynamic customer segments
- Sync of scores and metrics back into your CRM and support tools
MCP server and CLI
- An MCP endpoint served from your own environment
- Works with Claude, ChatGPT, Copilot, or an agent you build
- Sources and connections declared as code
Access and audit
- Single sign-on with your identity provider
- Role-based access control
- Audit history of data and definition changes
Bring your architecture diagram.
In 30 minutes we map RevOS onto your cloud, your lakehouse, and your security requirements.
AI inference
AI inference can stay on your premises too
The AI features in RevOS need a language model, and prompts can include query results. In a private deployment you decide where that model runs. RevOS does not route your prompts through a model account of its own.
Bring your own key
Your account with the model provider you approved
RevOS calls the model through your own API key and your own account, in the cloud and region you select.
- Prompts are covered by your contract and your data-processing terms
- Usage is billed to you and visible in your own provider console
- You rotate or revoke the key whenever you need to
On-premise endpoint
A model you host inside your own network
Point RevOS at a model endpoint that runs on your premises or in your private cloud, such as a self-hosted open-weight model.
- Prompts and responses stay inside your network
- No dependency on an external AI service
- Works in environments with restricted outbound access
For your AI agents
One governed layer for every agent in your company
Once RevOS runs in your environment, it becomes the safe place to point the AI tools your teams adopt. The MCP server ships with the deployment, so agents answer from certified metrics without your data leaving your network.
01
Any MCP client
Claude, ChatGPT, Copilot, and your own agents connect over the Model Context Protocol to one endpoint.
02
Inside your network
The MCP server runs with the rest of the deployment, so agent queries are answered next to the data.
03
Governed definitions
Agents read certified metrics from the Metric Store instead of writing SQL against raw tables.
04
Same roles, same audit
The access rules and audit history that apply to your people apply to every agent query.
The same platform, in your environment
Every part of RevOS is included in a private deployment.
Semantic Model
Metrics and dimensions defined as code, reviewed, and applied through the CLI.
Metric Store
One governed definition of every metric, with an owner, a history, and lineage.
Agentic Data Engineering
AI agents build and maintain the data layer through the CLI, with a human in the loop.
FAQ
Questions from security and platform teams
Our security and compliance program is published on the RevOS trust page.
Connectors, storage, modelling, metric computation, and the application all run inside your environment, and the data sits in your own lakehouse. The one path to review is AI inference, because prompts can include query results. You choose where it runs. Point RevOS at a model endpoint on your premises and prompts stay inside your network; bring your own key for an external provider and the prompts are covered by your agreement with that provider.
Yes. A private deployment supports two options. With bring your own key, RevOS calls the model provider you have approved through your own API key and account, so the prompts fall under your contract and the usage is billed to you. With an on-premise endpoint, RevOS calls a model you host inside your own network, and prompts and responses stay there. You can change the option later without redeploying the rest of the platform.
Bring your own key means the AI features authenticate to the model provider with a key from your account instead of a key owned by RevOS. Your security team keeps control of the provider, the region, the data-processing terms, and the spend, and can rotate or revoke the key at any time. RevOS stores the key in your deployment, with your other credentials.
RevOS uses the lakehouse or database you already run as its backend. That includes cloud warehouses such as Google BigQuery, Snowflake, Databricks, Microsoft Fabric, Amazon Redshift, and ClickHouse; query engines such as Amazon Athena, Trino, Presto, and DuckDB; and databases you host yourself, such as PostgreSQL, Microsoft SQL Server, MySQL, and Oracle. Data lands there, and the semantic model, metrics, scores, and segments are computed there. If you run a different engine, tell us which one and we will confirm the fit during scoping.
Yes. One deployment serves one organisation. It has its own application, its own database, and its own compute, and it shares none of them with other RevOS customers.
By default, nothing in your environment. The deployment needs no inbound connection from RevOS. We provide the software and your team runs it. When you want hands-on help, you grant our engineers access for that session and revoke it afterwards.
Yes. RevOS runs on Kubernetes, so it can be deployed in a public cloud account you own or in your own data center, next to a database you host such as PostgreSQL. We confirm the infrastructure requirements with your platform team before the deployment starts.
The RevOS MCP server is part of the deployment and runs inside your network. Agents connect to it and read governed metrics from the Metric Store, under the same roles as your users. Which model those agents use, and where it is hosted, is your decision. For the AI features inside RevOS, you can bring your own key or use an on-premise model endpoint.
Yes. Users sign in through your identity provider with single sign-on, and role-based access control decides what each person and each agent can see.
No. A private deployment includes the same connectors, Semantic Model, Metric Store, scoring, segments, Actions, CLI, and MCP server as the hosted product. The difference is where it runs and who operates it.
It stays where it is. Raw data, models, and metrics are standard tables in your own lakehouse, and metric definitions are code in your own repository. You keep both, and you can query the tables without RevOS.
RevOS publishes versioned releases, and your team applies them through your own change process. Nothing is pushed into your environment.
Private deployment is part of the Enterprise plan. The price depends on the scope of the deployment and the support level you need, and we agree it with you after the scoping call.
Yes. We complete security questionnaires and join calls with your security or vendor-risk team. Our security and compliance program is published at revos.trustshare.com. Reviews of a private deployment tend to be shorter, because the questions about where data is stored and who can reach it are answered by your own environment.
Keep your data where it is.
Walk through the architecture with our engineers, get your security questionnaire answered, and scope the deployment.
