Your Secure GenAI Platform

Narus is a generative AI chat portal that helps your teams work smarter, offering multiple large language model connections like GPT-4o, Claude, and Gemini. It’s designed for businesses that want to adopt AI securely with complete administrative controls, budget management, and oversight of AI usage.

Narus has been built for businesses that want to exploit the competitive advantage that AI offers without compromising security.

Narus allows IT managers to add and control GenAI tools in one place. It gives them clear information about how these tools are being used and lets them set security rules.

The Employee Portal lets staff get the most out of their GenAI tools, speeds up the learning process, and fosters collaboration around best practices leaving you with a best-fit model built by your own team.

Narus uses a dedicated Amazon Web Services (AWS) instance administered by Kolekti (part of The Adaptavist Group) for data storage. Employing robust encryption protocols and maintaining meticulous data segregation, we guarantee the confidentiality and integrity of your organisation's sensitive information, ensuring full compliance with industry-leading data privacy standards. See our Kolekti Data Protection Addendum (DPA) for further details.

Architecture

Yes, Narus is compliant with GDPR and ISO27001.

As a general rule, if an LLM is used through an API, the data is not used to train the model. However, some models offer the option to opt-in training, make sure your company has not actively requested it if you want models not to be trained with your data. Learn more about non-trained models that Narus is compatible with:

• OpenAI
• Google
• Anthropic

Mistral and Llama-2 (which are included in Narus by default) have both been configured not to train based on the inputed information. Furthermore, the data is not stored by the LLM provider, ensuring that your organization's information remains inaccessible and secure.

While traditional AI excels at analysing and categorising information by contrast, generative AI (or GenAI) can create new and original content in text, images and audio after being trained on huge datasets of the same kind of information. 

An LLM, or large language model, is a type of AI that has been trained on a massive dataset of text. This training allows LLMs to understand and generate human-like text, making them useful for a variety of tasks. Some examples of an LLM are ChatGPT, Gemini, and Claude.

Some tasks LLMs can help with:‍

An LLM, or large language model, is a type of AI that has been trained on a massive dataset of text. This training allows LLMs to understand and generate human-like text, making them useful for a variety of tasks. Some examples of an LLM are ChatGPT, Gemini, and Claude.

Some tasks LLMs can help with:‍

• Accessibility of Information: By interacting with an LLM, users can get explanations, summaries, or answers to questions based on the vast amount of knowledge the model has been trained on. This makes information more accessible to everyone, regardless of their background.
• Efficiency and Productivity: LLMs can automate tasks that involve language, such as summarising documents or meeting notes, drafting emails or reports, generating code, and saving time and effort for knowledge workers across various fields.
• Learning and Education: They can be used as educational tools, offering explanations on complex topics, assisting with language learning and even helping students brainstorm ideas for their projects.
• Understanding and Generating Language: LLMs understand queries or prompts in natural language (the way humans talk) and generate coherent, contextually relevant, and often indistinguishable responses from those a human might write.
• Innovation in Creative Fields: LLMs can assist in creative writing, generating ideas for stories, composing music lyrics, or even coming up with new recipes, serving as a collaborative partner for artists and creators.

Do note, however, that while LLMs can perform all these tasks, they can also at times produce incorrect or misleading information, generally referred to as hallucinations. LLMs are incredibly powerful and productive tools, but their output always needs to be verified.