Data Flow and Storage in Classic AI Chatbot

What is Classic AI Chatbot (CAC)?

CAC is an AI-powered assistant integrated into Simployer Classic HRM. CAC helps employees, managers, and HR administrators interact with HR data and company information through natural language. Users can ask questions about their own HR data, request and approve time off (future feature), access team insights, explore analytics – all through a conversational interface.

CAC is available through the web and mobile interface of Simployer Classic. The module is not enabled by default. A user with the owner/administrator role for the tenant can instruct Simployer to activate / deactivate it for the organisation or activate it themselves.

CAC connects to the same backend infrastructure and applies identical processing logic, security controls, and privacy safeguards as the Simployer Classic application. CAC does not change the categories of personal data processed, the sub-processor chain, or Simployer’s role as data processor.

How does data flow in CAC?

When a user interacts with CAC, the following sequence takes place:

1. The user sends a message through CAC's interface.

2. The user is authenticated and mapped to their Simployer identity. CAC determines the user’s role and permissions within the customer’s Simployer Classic tenant.

3. CAC interprets the user’s intent using Simployer's AI engine (Microsoft Azure OpenAI Services or Google Cloud Vertex AI; see the AI infrastructure table below for the current sub-processor list) and retrieves relevant information from the customer’s tenant data.

4. CAC generates a response and returns it to the user through the same channel. If the user requested an action (e.g., a leave request), CAC translates this into the appropriate HRM operation (future feature).

5. The response is displayed in CAC's interace. All AI outputs are assistive and require human review or confirmation for actions that modify data (future feature).

All processing in steps 2–4 occurs within Simployer’s infrastructure, hosted exclusively in the EU/EEA.

Where is data stored?

CAC uses two types of storage within Simployer’s infrastructure:

Storage	What is stored	Location
Simployer HRM database	All customer data processed by CAC is in the same database as Simployer Classic HRM. This includes any data CAC reads or writes as part of HRM operations.	Microsoft Azure and GCP, EU/EEA
LiteLLM Postgres database	Conversation logs. Log data is anonymised and does not contain identifiable customer data.	Microsoft Azure, EU/EEA

 

Data is not mixed between customers. Each customer’s data is strictly isolated within their own tenant.

Important: Customer data is not used to train or modify the underlying AI models. User feedback (thumbs up or thumbs down and potential comment) is collected solely by Simployer for feedback and statistical purposes to improve the solution.

AI infrastructure

CAC’s AI processing is powered by the following components:

Component	Provider	Purpose	Location
Large Language Model	Microsoft Azure OpenAI Services	Natural language understanding, intent interpretation, response generation	Microsoft Azure, EU/EEA (Microsoft Ireland)
Large Language Model	Google Cloud Vertex AI	Natural language understanding, intent interpretation, response generation	Google Cloud, EU/EEA (Google Ireland Ltd)
Application APIs	Simployer (internal)	Authentication, data retrieval, HRM operations	Microsoft Azure, GCP, EU/EEA
Data anonymization	Simployer (using Presidio internally)	Data anonymization	Microsoft Azure, EU/EEA
Rate limiting and cost tracking	Simployer (using LiteLLM internally)	API proxy, observability / cost tracking	Microsoft Azure, EU/EEA

 

The sub-processors for AI inference are Microsoft Ireland (Azure OpenAI) and Google Ireland Ltd (Vertex AI). See the sub-processors page for the complete list.

What personal data does CAC process?

CAC can access and process the same categories of personal data available in Simployer Classic HRM, limited by the individual user’s role and permissions. Depending on the customer’s configuration, this may include:

• Contact information (name, email, phone number)

• National identification numbers

• Employment information (position, department, manager, start date)

• Absence and time management data (future feature)

• Special categories of personal data (e.g., health-related absence data, where applicable) (future feature)

CAC does not have broader access than the authenticated user. A user can only retrieve information they are already authorised to see in Simployer Classic.

Delivery channel

CAC is only available through the web-application channel (a web browser). The user interacts with CAC directly within the Simployer Classic interface. Chat logs are stored and managed within Simployer’s infrastructure. Log retention are 90 days. only dedicated technical Simployer personnel can access chat logs in order to provide technical support and error handling.

Security and privacy safeguards

The following safeguards apply to all CAC interactions:

• Role-based access control: CAC enforces the same role and permission model as Simployer Classic. Users only receive information they are authorised to access.

• Tenant isolation: Data is strictly isolated between customer tenants. CAC cannot access data belonging to other customers.

• No model training: Customer data is not used to train, fine-tune, or modify the underlying AI models.

• EU/EEA processing: All AI processing occurs exclusively within the EU/EEA.

• Encryption in transit: All communication between the end users browser and Simployer’s backend is encrypted using TLS/HTTPS.

• Assistive AI only: CAC does not make autonomous decisions. All outputs are assistive and require human review or confirmation.

• Audit logging: Interactions with CAC are logged for security and compliance purposes.

Related pages

• Processing Activities in Simployer Classic

• Sub-processors

• Data Processing Agreement

• Use of AI in Simployer HRM Systems

• Use of AI in Classic Handbooks