Grounded answers
Responses are built from the content and system data available to the agent. When a source does not provide an answer, Chatevo does not substitute an invented value.
Trust & architecture
Designed so your agent can be trusted with real business information.
Chatevo combines knowledge retrieval with live system access. The controls around those connections determine what the agent can use, what it can return, and where sensitive credentials remain.
Chatevo is pre-launch. This page describes the product controls and practices currently implemented. It does not represent a security certification.
Controlled access
Credentials are used server-side and are not exposed to the website widget. Deployments can also be restricted to approved domains.
Isolated customer data
Knowledge and conversation data are associated with the organisation that owns them and are not shared across accounts.
Response architecture
What happens before an answer reaches the customer.
Trust starts with understanding where a response comes from. Chatevo separates the customer request, retrieval, connected system response, and final answer rather than treating them as one opaque model call.
01
Stage
Customer question
The customer asks for information or an action through your agent.
02
Stage
Relevant context
Chatevo determines which knowledge or connected system is relevant to the request.
03
Stage
Source response
The agent retrieves the relevant content or waits for the connected system to return its response.
04
Stage
Final answer
The response is assembled from the information actually retrieved.
The model is not the source of truth.
When a response depends on live business data, Chatevo uses the value returned by the connected system. When it depends on your knowledge, it uses the retrieved source content. The final response is assembled from those inputs.
Product controls
Controls built into the response and connection layer.
These are implementation boundaries within the product rather than instructions provided to the model.
Response assembly
Structured values come from source data
Prices, inventory, order status, delivery dates, and other structured values are rendered from fields returned by your connected systems. The model does not generate those values itself.
Source attribution
Knowledge answers retain their source
When an answer comes from your knowledge base, the response can identify the document and source context used to support it.
Scoped retrieval
Retrieval stays within your knowledge
Knowledge retrieval is scoped to the documents and pages you provide. Chatevo does not use the open internet as an unbounded fallback for answers.
Fallback behaviour
Unknown information stays unknown
When available content and connected systems cannot answer a request, the agent can state that the information is unavailable and route the customer to your team.
Server-side credentials
Credentials remain server-side
API keys, bearer tokens, OAuth refresh tokens, and HMAC secrets are stored and used on the server. They are not exposed through the embedded client.
Domain allowlist
Deployments can be domain-restricted
Each deployment can specify which domains are allowed to load the assistant, reducing the risk of unauthorised embedding and unintended usage.
Data handling
How your business data is handled.
Chatevo keeps knowledge, conversations, and connected system credentials within the boundaries of the organisation using them.
Model training
Your documents and customer conversations are not used to train public AI models.
Content storage
Documents are stored in object storage, while their content is indexed for semantic retrieval. Both are associated with your organisation.
Deletion
Removing content removes it from the corresponding knowledge index. Organisation-level deletion removes the associated organisation data.
Operational access
Administrative access is limited to operational needs, and administrative actions are recorded through audit logging.
Transparency
What Chatevo does not claim today.
Chatevo does not currently hold a security certification such as ISO 27001, and this site does not display certification badges that imply otherwise. The product currently provides the controls and practices described on this page, including encryption, scoped retrieval, server-side credential handling, domain restrictions, and audit logging.
As the product matures, additional security controls and independently audited certifications can be added here with their exact scope and supporting details.
Have a security or architecture question?
Share the requirement, integration pattern, or control you need to evaluate. We can walk through how Chatevo handles it before you move into implementation.