Talk to Your Business Systems
Admin Panel
| Username | Display Name | Role | Credentials | Actions |
|---|
Map each user to their identity in each enterprise system. This controls what data they can access via RBAC (role-based access control).
API keys allow programmatic access to the chat API. Each key inherits the RBAC permissions of the user it's assigned to.
| User | Name | Key | Created | Last Used | Actions |
|---|
Authentication: Include the key in the X-API-Key header or as Authorization: Bearer ik_...
Chat endpoint: POST /api/chat with {"message": "your question"}
Example:
curl -X POST http://localhost:8080/api/chat \
-H "Content-Type: application/json" \
-H "X-API-Key: ik_your_key_here" \
-d '{"message": "How many open work orders are there?"}'
Full API reference: Open Interactive API Docs (Swagger UI)
Teach the chatbot new queries by adding custom patterns. These extend the built-in knowledge for your specific environment.
Paste your license key (INTRP-xxx) or JSON below, or load from a file.
Enable verbose logging for troubleshooting. Debug mode generates larger log files — disable when not needed.
When enabled, anonymous aggregate metrics are sent daily. Every payload is logged below so you can verify exactly what is transmitted.
Every telemetry payload is logged here before it is sent. This file also exists on disk at /app/data/telemetry_audit.jsonl.
Every query executed against your connected systems is logged here — SQL for EBS/PeopleSoft, API calls for Maximo. This file also exists on disk at /app/data/query_audit.jsonl.
Download a diagnostic package to share with Interpretos support. Includes system info, connectivity test results, error summary, and anonymised audit logs.
Privacy: No credentials, SQL text, conversation content, or usernames are included.
Main AI model for answering user questions and generating code.
Leave blank to keep the existing key
Provider-specific params. Gemini: {"thinking_budget": 0}
OpenAI: {"reasoning": {"effort": "none"}}
Picks which knowledge pattern to use for each query. Should be fast and cheap.
Auto-upgrade model for complex queries. When the system detects a hard question, it routes to this model instead of the Chat LLM.
Leave blank to keep the existing password
SSH + sqlplus connection to PeopleSoft HCM database.
Leave blank to keep the existing key
Leave blank to keep the existing password
When enabled (and at least two integrations are configured), questions that span systems — e.g. "find this work order's PO and check its status in EBS" — are answered by querying each system separately and combining the results. Single-system questions are unaffected. Each cross-system step runs with the asking user's own credentials and access scope.
How long a single question may run, and how much context a conversation may use before it is stopped. Raise these for heavy analytical questions; leave a field blank to use the default.
How long a single question may run before the user is told it timed out.
Total tokens a conversation may use before the user is asked to start a new chat. The AI model's own context window is the real limit and may bite first.
Default context for all users of an integration (max 1000 chars). These are injected as defaults the AI will follow unless the user overrides them in their message.
Suggestion chips shown to users when they start a new conversation. Up to 6 per integration.
Re-run the guided setup to change your AI provider or database connections. Your license and user accounts will be preserved.