Saklam is an AI privacy layer from Germany-based Portalix UG. Its core goal is to help lawyers, doctors, tax advisors, and other professionals subject to §203 StGB confidentiality obligations use large language models safely. Before prompts are sent to models such as OpenAI, Anthropic, or Mistral, Saklam automatically replaces sensitive information—such as names, addresses, case numbers, tax IDs, insurance numbers, IBANs, and email addresses—with tokens, then restores the original data after the AI responds.
Saklam is not focused on building its own large language model. Its emphasis is on “redaction + proxying + review.” The web version supports in-browser processing of text as well as PDF, Word, and Excel files, and includes a Review Mode so users can see what is actually being sent to the AI. The desktop version supports macOS, Windows, and Linux, with an emphasis on 100% local detection; iOS and Android mobile apps are currently in Beta. Developers can use the Mask API or an OpenAI-compatible Proxy to embed PII masking into existing LLM pipelines. The enterprise Bridge edition can be deployed locally via Docker and integrated with Claude, GPT, Gemini, Azure OpenAI, AWS Bedrock, Mistral, or self-hosted Ollama/vLLM.
The standard plan costs €79/month plus VAT and includes a €79 AI Budget, which can be used across different models, document masking, and Review Mode. If usage runs short, top-ups start from €10, and the plan can be canceled at any time. Bridge offers a 30-day free Pilot; the full Pro edition costs €299/employee/year, while Enterprise costs €499/employee/year. Payments are handled through Stripe, with support for credit cards, SEPA, and other methods.
Its strengths are a very clear positioning and a design built around German professional confidentiality, GDPR, Zero-Knowledge principles, and EU hosting. It also offers a broad range of integration options, covering everything from regular chat to APIs, proxies, and local deployment. The limitations are also clear: PII detection is not guaranteed to be 100% complete, and the AGB requires users to review results themselves; some features are still marked as Beta or coming soon; and Bridge is closed-source proprietary software. Chinese-language support is not mentioned.
Saklam is suitable for high-sensitivity data scenarios in Germany and the EU, including legal, medical, tax, consulting, and software outsourcing use cases. It is also a fit for development teams that need unified redaction before LLM calls. The materials do not state how well it works from China; network connectivity and the availability of Stripe payments would need to be tested. For compliance within mainland China, an alternative setup combining locally deployed redaction components with domestic large model APIs may be worth considering.
⚠ This review is compiled from public sources and does not constitute a purchase recommendation. Verify all facts on the vendor's official site. Verify on saklam.com official site.
saklam.com is an Unknown AI Apps provider. TG4G tracks its product information, an overall rating of 8.0/10, and a China-accessibility score of Workable. Click "Visit Official Site" to reach saklam.com directly.