AIMAX^®
- Your central enterprise AI platform

AIMAX® includes granular access control lists (ACLs) for agents, data and interfaces. 

You determine:

• who is permitted to use which agents
• who is granted access to which data via artificial intelligence
• who is permitted to access which interfaces using the agents
• who is granted access to which AI model

You can centrally determine which channels are available to your staff (Microsoft Teams, WebUI, Outlook 365, Zendesk, etc.) and which features these provide.

You can also specify who is authorised to use which AI models and, for which models, for example, separate manual confirmation is still required.

Group the AI models by area of application as a practical alternative to cryptic model names.

You can also manage these and other settings centrally on a client-specific basis.

It is not uncommon for new AI models to behave slightly differently from their predecessors. As a rule, this behaviour can be influenced by adjusting the system prompt. To this end, AIMAX® allows you to define different system prompts for each AI model or model family.

This allows you to ‘normalise’ the behaviour of the models and thus ensure consistent result quality.

Furthermore, the system prompts can also be customised on a channel-by-channel basis. This enables AIMAX®, for example, to cater to the different characteristics or features of the channels and thus make better use of their capabilities.

The integrated knowledge cache can store information in a wide variety of formats. As well as PDFs, Office documents, images and numerous other file formats, this also includes the contents of email inboxes, intranet pages and, for example, FAQ collections.

This knowledge is made available to your staff when they use the AI – naturally, in accordance with their respective roles and access rights.

AIMAX® automatically keeps this knowledge up to date at all times. This avoids duplicate data storage.

The quality of the results for a task handed over to AI depends on more than just the prompt. Naturally, you must also provide the AI with the information it needs to answer the question.

However, if you provide the AI with too much irrelevant information, the result will suffer as a consequence.

The context engineering functionality integrated into AIMAX® reduces the context to the essentials, thereby increasing the reliability of good results.

Define any agents or experts who will carry out the tasks assigned to them on your behalf.

Assign a specific role or behaviour to the agent for this purpose, set clear boundaries and define when the agent should be deployed.

Finally, you determine the tools available to the agent and grant access to the specific specialist knowledge.

AIMAX® provides you with a tool builder featuring around 500 modules and connectors, ready to use straight out of the box, for orchestrating complex workflows. For example, you can easily have AI results checked by a second AI (Second Revisor) or combine generative AI with deterministic RPA processes.

You decide for yourself how, upon which event or at what time such a workflow should be triggered (event-based triggering – e.g. based on time, email receipt, data change, etc.).

MCP stands for the Model Context Protocol, which connects AI models and AI agents to external data sources and tools.

AIMAX® includes a fully-fledged MCP client, enabling it to utilise the relevant data sources and tools.

Both the older SSE method and Streamable HTTP are supported for communication.

Furthermore, you can centrally define which of the tools provided by the MCP server are authorised for use and which user rights are required for them.

As AI is integrated into an ever-increasing number of business processes, it is simultaneously becoming an increasingly business-critical resource. This means that its availability is also becoming ever more important.

For this reason, AIMAX® incorporates a mechanism for load balancing and failover. This is achieved in the software architecture through the ability to utilise AI workers within a computing cluster, thereby enabling an uninterrupted handover in the event of a component failure.

Furthermore, this technology also enables the use of a wide variety of AI hardware in a networked configuration.

The central administration is complemented by the option to integrate with your company’s Active Directory. User accounts and company-wide security policies are thus still managed in their existing location and synchronised with AIMAX®.

Despite the centralised AI solution and rights management, every user in AIMAX® naturally has their own personalised context and can decide for themselves in which language, tone or even style they wish to interact with the AI.

With the integrated human-in-the-loop mode, you can specify which processes require manual checking and review of the AI results.

AIMAX® also allows you to correct the results directly where necessary to ensure a smooth workflow.

Furthermore, when using cloud-based AI models, prior approval from the user can be required.

Security remains the top priority, even when connecting additional systems via the OpenAI-compliant API.

AIMAX® therefore allows for the management of different API accesses, including the specification of the user context in which they operate. Furthermore, it is possible to impose additional restrictions on the tools and AI models that can be used.

The aim: to reduce external access rights to the absolute minimum in order to ensure the highest possible level of internal security.

AIMAX® enables the definition of ontologies (entities and relations) to enhance contextual understanding and uncover implicit information – comparable to the knowledge between the lines.

The resulting higher quality of AI responses also leads to fewer hallucinations.

The integrated decision log allows you to determine at any time on the basis of what information, in what context, and by which member of staff or agent a decision was made.

You can also keep track of the active user sessions on your stack.

AIMAX® is not limited to the specialist knowledge included in the respective model and assigned in addition.

The integrated browser agent can also search the internet for further information. If necessary, it uses Google to find a page and is also able to dismiss cookie banners.

The REST API connector integrated into AIMAX® enables external applications to be connected via the REST interface.

With just a few clicks and without any programming knowledge, you can define the REST endpoint of the application to be connected in AIMAX® and specify whether the human-in-the-loop function should be active for this. Done!

In addition to the integrated MCP client, AIMAX® also includes its own MCP server. This allows you to make the internally defined functionalities available to external AI agents if required.

As part of our mission to make the power of artificial intelligence accessible to businesses in a simple and sustainable way, we have put numerous hardware solutions through their paces. The solution identified as part of this process, based on the Apple Mac Studio, is characterised by very high performance, maximum efficiency and excellent value for money.

For running local AI in small to medium-sized enterprises, we recommend running AIMAX® on this platform. In combination with our software, we have already been awarded a sustainability prize for this.

Of course, you are free to use AIMAX® on an alternative platform as well. We would be happy to assist you in making your decision.