With Neo, we not only have developed a digital assistant for business, but also powerful back-end systems and tools that allow you to implement an individualized AI-assistant for your customers.
Powered by our integrational framework, you're able to connect various Enterprise systems and tools to Neo. Our Drag-and-Drop workflow engine allows you to easily create conversations and build UI-elements, and our Natural-Language-Understanding platform enables you to create, customize and train speech models.
With Neo as AI-assistant, interaction with software is not only simple and intuitive for end users: Implementing custom bots and assistants is a lot more efficient – and trust us, we Germans do love efficiency – thanks to our integrational framework and drag-and-drop interfaces.
Wit our client application for mobile (iOS and Android) and Desktop (Windows, macOS and Linux), users are able to interact with Neo to access and query their IT-systems.
Creating and editing intents is a breeze with our Natural-Language-Understanding platform and requires absolutely no coding skills.
Our Neo Flow Engine allows you to build custom workflows using a Drag-and-Drop UI. You're able to integrate third-party-systems in 18 programming languages for maximum efficiency (SDKs available).
Particularly in the Enterprise context, IT products must be able to be adapted to the company's unique requirements: One often encounters a heterogeneous system landscape; in some cases, self-developed systems are in use, and legacy-systems must also be taken into account. Thus, a project quickly becomes an incomprehensibly complex task for both, the developer and the end user. We know these problems from our day-to-day work - and have created a toolset for us that addresses these very aspects: Our Neohelden Toolbox allows implementation of individual AI-assistants with custom system integrations to third-party systems in complex deployment scenarios.
Our tool suite is based on three fundamental pillars: (1) an easily configurable, flexible and highly innovative user interface (the "bot" for end users), (2) a natural language understanding platform, and a (3) workflow engine with a drag-and-drop interface allowing integrations in 18 programming languages in order to integrate existing IT systems. On the one hand, this toolset greatly simplifies processes for end users and makes them accessible via a platform on all end devices via voice and text; on the other hand, the integration-effort is reduced to the essentials in the best possible way. In addition, drag-and-drop configurations allow fast prototyping cycles and early involvement of end users for maximum acceptance.
Chatbots and digital assistants have been part of everyday life through their ubiquity at home and on our mobile devices: From Amazon Alexa to Apple Siri, Google Assistant, Facebook messenger chatbots and telegram bots. In the enterprise context, however, data protection is a critical aspect. Hence, building on top of well-known B2C platforms is not feasible when it comes to data sovereignty.
Therefore, we have decided to create our own application that meets the needs of the enterprise environment: Our Neo application can be installed on all systems already in use. Mobile devices and tablets are supported on iOS and Android, desktop devices and Notebooks with Microsoft Windows, macOS and Linux. No additional hardware is required, and no data is flowing through third-party channels. Nevertheless, we build on top of existing industry standards and can integrate flexible and standardized visual formats, such as Microsoft Adaptive Cards, into the bot. Thus, we combine security and data-control with high flexibility and standardization.
When it comes to chatbots and voice or text-based agents, the language model is essential for meaningful and productive uses of the bot. With our Neo Conversation Engine, so-called "intents" (i.e. commands) can be mapped via example Utterances (possible statements of the users) in a language model – that is not keyword-based, but contextually translates the users' input into their actual intentions. Training and maintenance of the language models can take place via a user interface for which no programming knowledge is necessary. As an alternative to custom language models, which can be trained and used on-premises, Natural Language Understanding modules of renowned cloud providers such as AWS or Google can be connected. However, this does not allow on-premise operation but could increase the number of languages supported.
Interaction with our AI Assistant Neo started out being text-based but is now also accessible via voice. We think of voice as an additional layer, where voice-input is transcribed to text, mostly using cloud providers (e.g. Bing Speech API), and then interpreted into the actual intent via the Neo Conversation Engine and the underlying speech model. Replies from our AI-assistant come in form of both visual content and voice explanation, to further enrich information.
Especially in B2B, chatbots and virtual agents are less tasked with querying FAQ systems, but often involve handling entire processes – for example, room bookings, scheduling, budget inquiries or travel expense reportings. Often, these processes are highly individualized and may involve a variety of third-party systems for sub-aspects of the process. Using the Neo Flow Engine and our Integrational Framework, third-party systems can be connected in 18 programming languages. This minimizes implementational efforts, since existing libraries and frameworks can be used, or at least reduces the effort to a feasible scope so that no new skills in unknown programming languages and development environments are required.
Getting users involved early on is a high priority in Enterprise software projects – which essentially are Change Management projects. Hence, for quick prototyping, workflows can be modeled via a drag-and-drop interface. Entire processes and conversations can be modeled in real time and tested directly with end-users to allow maximum acceptance and usability. This enables an efficient mapping of business logic – without implementation effort. As soon as a workflow has been tested and found to be useful by end users, the implementation can optionally be transformed in a high-performance, lean programming language, such as Go by Google, whereby the business logic can be adopted from the flow engine.