You decide when and how your AI communicates with the internet, using domain whitelisting, DLP inspection, and complete logging.
wikipedia.org
Knowledge source
arxiv.org
Research
openai.com
External AI service
google.com/search
Search engine
intern.firma.de
Intranet
unknown-tracker.io
Unknown
Outgoing requests are inspected for sensitive data. No company data leaves the network.
Transparency notice: this demo is a simulation using prepared sample data and does not show output from a live AI model. Labelled in line with Art. 50 of Regulation (EU) 2024/1689.
Define exactly which domains the AI can reach. All unapproved targets are automatically blocked.
Every outgoing request is inspected for sensitive company data. Personal data, contract content, and trade secrets are automatically filtered.
All inbound and outbound connections are recorded, with user, time, and requested source.
All internet requests are routed through your corporate proxy and therefore through your existing network security.
ThinkLocAI runs completely without internet. Internet access is optional and must be explicitly enabled and configured.
Dashboard for outgoing traffic, with alerts on unusual patterns or attempts to reach unapproved targets.
The AI can search approved knowledge sources such as Wikipedia or industry portals without company data leaving the network.
Financial institutions and government agencies can granularly control internet access and fully document all data flows.
Run ThinkLocAI in air-gap mode and enable individual sources selectively. Every access is logged.
We show you how to approve domains, create DLP rules, and review the logs.
