Where your data lives.And where it goes.
Your files, questions and answers all have a home. See how we set up Exemplary on your own servers or in the cloud, how it connects to AI models, and what happens to your information along the way.
You set the boundaries. We handle the AI setup.
Where is my data stored?
Your product data is stored where your Exemplary deployment runs: on your own servers, in your company’s private cloud, or in a cloud environment managed by us. Where the AI model runs is a separate choice. If you connect an external model, the content included in its requests goes to that provider.
See what stays. See what moves.
Choose where the product runs, then where the model runs. The drawing shows how those two choices change the journey of a question and the information used to answer it.
1. Where is the product installed?
2. Where does the AI model run?
Your team asks a questionFor example: “What is our travel policy?”
Files & search data
Stored copies and a search index help find the right information.
Exemplary AI
Checks access and prepares the question with relevant information.
AI model
The model runs on computing hardware inside your network.
Saved conversations and generated files are stored in this deployment too.
The answer returns to your team in Exemplary. Document-based answers can link to their sources. Sending a result to another app is a separate connection.
Storage and model processing are in the same environment.
With local models and local services, the files and questions used by this workflow stay inside your network.
We install and test the AI. Your team runs the infrastructure after handover.
On-premises
Your own servers, at your office or data centre.
Your private cloud
Servers in a cloud account controlled by your organisation.
Managed cloud
Dedicated cloud infrastructure operated by Exemplary for your organisation.
From a machine to a product your team can use.
You do not need to piece the AI together yourself. We work through the setup with your team, from choosing capacity to testing the finished workflow.
Start with the right machine or cloud setup.
Before choosing a computer, we look at the work: which files you use, which AI models you need and how many tasks will run at once. That tells us how much computing power, memory and storage the setup needs.
On your premises
We size the setup with your team around approved hardware. Your IT team provides the machine, power and network; we prepare the AI installation for it.
In the cloud
We plan the cloud capacity and hosting location. Your organisation controls a private cloud account; Exemplary operates a managed cloud deployment.
How we help: You explain the work. We help translate it into a practical setup, without asking you to choose AI hardware on your own.
Install the product and connect the models.
We set up the workspace your team will use, the storage behind it and the AI models that will answer requests. An AI model is the part that reads the information it receives and produces a response.
On your premises
Local language and search models run on your hardware. A fully offline setup uses local services and approved offline installation packages.
In the cloud
The product runs in the chosen cloud environment. Models can run there too, or requests can go to an external model service you approve.
How we help: We handle the AI installation and model connections. We make the destination of each connection clear before it is used.
Bring in the files and tools your team needs.
You choose the documents, folders and business systems the product can use. Uploading or syncing documents can create copies in the deployment; it does not simply point at the originals.
On your premises
Uploaded files and imported copies are stored in your deployment. Connecting to an online app still needs a connection to that app.
In the cloud
Uploaded files and imported copies are stored in the cloud deployment. The original records in a connected business system remain in that system.
How we help: We help connect the right sources and set access for the people and agents that need them.
Make those files ready to search.
The system reads the files, extracts text and creates a search index: a set of references that helps it find relevant passages later. These extra copies of information need the same care as the original files.
On your premises
With local document processing and search models, the text and search index are created and stored inside your environment.
In the cloud
The text and search index are stored in the cloud deployment. If an external service processes a file or creates search data, the content it needs is sent to that service.
How we help: We set up document processing and search together, so the location of the original file is only one part of the data map.
Ask a question and get a useful answer.
When someone asks a question, the product checks their access, finds relevant information and sends the question and supporting content to the selected model. The response comes back to the product, with source references for answers based on your documents.
On your premises
A local model reads that content inside your network. If you choose an external model, the request content leaves your network to be processed there.
In the cloud
The same process runs in the cloud. An external model provider is a separate destination, even when the product is already cloud-hosted.
How we help: We connect the model to the workflow and test it with your team. You review whether the answers are useful and which actions need approval.
Keep track of what happens after the answer.
The data journey continues after a reply. Saved chats, generated files, activity records, backups and anything sent to a connected app can all hold information from the task.
On your premises
Your team operates the deployment after handover, including infrastructure and backup arrangements. We provide the operating instructions; ongoing support is agreed separately.
In the cloud
Your team operates a private cloud deployment. In managed cloud, Exemplary operates the service under the agreed scope, including the agreed backup and support arrangements.
How we help: We test the setup, explain the operating responsibilities and agree how support and changes will be handled.
Using a file to answer a question does not train the model.
To answer from a company policy, the product finds relevant passages and includes them with the question. The model uses that information to write its response. This is different from changing the model by training it on your documents.
Exemplary does not use your enterprise data to train its models. If you choose an external model service, we also need to consider that service’s data terms, processing location and how long it keeps requests.
“Not used for training” and “never leaves your environment” mean different things. We make both clear for the setup you choose.
Explore model choicesWhat gets stored along the way?
A clear data map includes the information created while the product works. We account for these records as part of the deployment discussion.
- Files and imported copies
- Documents you upload and records copied from connected sources live in the deployment’s storage. Connecting a source can mean storing another copy.
- Extracted text and search data
- Text read from files and the search index live alongside the product’s data services. They can contain or represent sensitive information too.
- Conversations and results
- Saved questions, answers and generated files live in the deployment. A file you download creates a copy on your device; a result sent to another app creates data there.
- Activity and support records
- Activity records can show who accessed a file or used a tool. Diagnostic material shared for support is another data transfer to review, including who receives it.
- Backups and older versions
- Backup locations and deletion arrangements depend on the deployment plan. Removing a file from the product does not, by itself, establish when backup copies or external copies are removed.
Leave the AI setup to us. Know who runs it next.
We install and test the product with you. Day-to-day responsibility depends on whether your organisation runs the infrastructure or chooses our managed cloud service.
| Responsibility | On-premises or your private cloud | Exemplary-managed cloud |
|---|---|---|
| Choose where data may go | Your organisation sets the requirements. We design the AI setup around them. | Your organisation sets the requirements. We agree the hosting and processing locations with you. |
| Install and connect the AI | Exemplary installs, integrates, tests and hands over the product. | Exemplary sets up and operates the managed deployment. |
| Run the infrastructure | Your team runs the servers or private cloud after handover. Additional support is agreed separately. | Exemplary runs the cloud deployment under the agreed service scope. |
| Manage access and allowed actions | Your administrators decide who can use each resource. We help configure and test those permissions. | Your administrators decide who can use each resource. We help configure and test those permissions. |
| Set backup, deletion and support arrangements | Your team owns the operating policy. We clarify what must be handled in the product, infrastructure and connected services. | We agree the policy and service responsibilities with you, including backup location and support access. |
Choose where data may go
- On-premises or your private cloud
- Your organisation sets the requirements. We design the AI setup around them.
- Exemplary-managed cloud
- Your organisation sets the requirements. We agree the hosting and processing locations with you.
Install and connect the AI
- On-premises or your private cloud
- Exemplary installs, integrates, tests and hands over the product.
- Exemplary-managed cloud
- Exemplary sets up and operates the managed deployment.
Run the infrastructure
- On-premises or your private cloud
- Your team runs the servers or private cloud after handover. Additional support is agreed separately.
- Exemplary-managed cloud
- Exemplary runs the cloud deployment under the agreed service scope.
Manage access and allowed actions
- On-premises or your private cloud
- Your administrators decide who can use each resource. We help configure and test those permissions.
- Exemplary-managed cloud
- Your administrators decide who can use each resource. We help configure and test those permissions.
Set backup, deletion and support arrangements
- On-premises or your private cloud
- Your team owns the operating policy. We clarify what must be handled in the product, infrastructure and connected services.
- Exemplary-managed cloud
- We agree the policy and service responsibilities with you, including backup location and support access.
The questions behind “Where is our data?”
Does on-premises mean our data never leaves the building?
Does connecting our documents train an AI model on them?
Can we use an external model while keeping the product on our own servers?
Can every AI feature run without the internet?
Who chooses the cloud region and backup location?
Do we need to work out all of the technical details ourselves?
Follow the idea further.
Understand the concept, explore the evidence or see it applied to a task.
Grounded answers
A grounded answer is based on relevant source information supplied to the model. In a company assistant, that often means finding permitted documents first, then answering with references the reader can check.
Private AI
Private AI is an arrangement in which access to AI systems and the data they use is controlled for an organisation. Where the application, files and model run—and what leaves that boundary—determines what “private” means in practice.
Find what you need next.
Bring your data requirements. We’ll handle the AI setup.
Tell us what your team needs to do and where information is allowed to go. We’ll help plan the hardware or cloud environment, connect the right models and show you how the pieces fit together.