-
Process Information like Humans Do
-
Perceptor
Perceptor automatically extracts content from scans and digital documents. Maximum precision, traceability, and security in operation.
-
Trace
Trace finds the right information in huge document archives. This means that information is available with a single query. Reliable. Fast. Secure.
Our Customers
Products
PERCEPTOR
Universally applicable
Perceptor can be used both for recognizing document types with freely selectable taxonomy and for recognizing simple (e.g., header data) and complex information (e.g., tables) from documents. It supports multi-page documents, variations in layout, and different languages.
Fast & accurate
Perceptor offers a seamless transition from zero-shot integration (no training required, get started right away) to fine-tuning (flavor training with maximum recognition performance without hallucinations) via a single interface. Perceptor can be used for both real-time extraction (<4s/page) and batch processing. Perceptor provides page references with the answers for traceability of the information source.
Access via API
Perceptor can be easily accessed via REST API. This allows Perceptor to be easily integrated into a wide variety of system landscapes. Simply send a document with the list of information you are looking for. Perceptor returns the answers it finds in JSON format.
Operation: Cloud to on-premises
Perceptor is developed by TamedAI GmbH in Germany. Without external model dependency, Perceptor can be operated directly as a service in the cloud or in a cloud of your choice (up to your own data center).
TRACE
Fast & Accurate
Have you ever asked yourself: where is the latest project report? Where can I find information about the new security policy for authentication? Today, these questions result in a high level of communication effort within an organization.
Current solutions fail due to the heterogeneity of documents. For example, how can information from arbitrarily complex graphics be understood without ending up in hallucinations?
Trace solves this situation by making shared information easily available to the team. Simple questions quickly find and provide the information you are looking for. Trace focuses on searching for the right places within the knowledge base. Trace can not only access document fragments, but even read entire documents!
Intuitive & Secure
Trace comes with an intuitive web UI that not only quickly tracks down the information you need, but also allows you to access the information itself without switching to other applications. Fine-grained access control ensures that confidential information remains accessible only to specific individuals.
Operation
Trace is developed and operated by TamedAI GmbH in Germany. In addition to its use as a classic SaaS solution, Trace can also be used in hybrid mode (data management locally in the company, LLM services in the cloud) or completely on the intranet (on-premise).
The Company
TamedAI was founded as a start-up in 2019. Since then, we have begun developing and training small to large language models with the aim of bringing these models into productive use cases.
We are a team of highly specialized machine learning and software engineers working on products to make the exchange and access of information more efficient.
As a university spin-off, we remain closely connected to current research and are involved in several active research projects to further enhance the performance of our products.
Get In Touch
Interested?
Just drop us a message. We meet for a coffee and then we will wow you with what Perceptor and Trace can do.