More intelligence for your COGNIGY chatbot

Your Cognigy chatbot sometimes doesn’t really know what to do and either doesn’t provide an answer or doesn’t provide an appropriate answer to your customers’ questions?

What do you think about us helping you make your Cognigy chatbot smarter and making sure the chatbot offers additional information if it can’t find an answer to a customer query? As well as significantly reduce the maintenance and training effort for your chatbot at the same time?

Chatbotflow Cognigy to iFinder

We as kompaktwerk GmbH are specialized in enterprise AI-based search. We implement solutions for data integration and aggregation, as well as content enrichment.

In other words: We make information in large data sets quickly findable!

Preparing your chatbot for unforeseen questions and providing more in-depth information to the user … all delivered by our AI-based search and knowledge management solution iFinder.

iFinder Architecture

We have developed a powerful search extension for interactive question-answer chatbots from Cognigy. Through our solution, your chatbot can access enterprise-wide information quickly and efficiently. This enables you to optimize your chatbot dialogs and significantly improve user satisfaction.

How exactly does our solution work:

We extend your existing chatbot solution with a so-called Natural Language Question Answering module (NLQA).

This ensures that pre-formulated answers are filtered out of your company data by our iFinder and fed into the flow.

The iFinder is an AI-based knowledge management solution that extracts terms, texts, keywords and corresponding questions from different data sources and delivers suitable results and hits.
This is done through a small content hit list that we intelligently filter from your existing in-house data sources.

Processing is done by augmenting answers to your natural language search queries using Natural Language Processing (NLP).

That is, we “augment” your existing chatbot or voicebot info source.
Thus, we extend your BOT system with a “NLQA Module” -Natural-Language-Question-Answering- to which we can connect any info source via a Rest API, e.g. also your existing systems like M365, Confluence, Sharepoint, JIRA, Wikis, Salesforce etc.
The feed can be done via a JSON interface, Rest API or even web browser plugin and is injected into the flow, e.g. via a “More-Button”.

Our application searches not only in your chatbot database, but also in other data sources, e.g. Confluence, Sharepoint, other CMS, possibly also voice or chat transcripts.

With our Question Answering Machine, voice control of the search is also possible. The iFinder comes with an optional speech-to-search function. Of course, everything also runs onPremise or as a SaaS service in the cloud.

Feel free to take a look at our digital showroom for more details:
iFinder – more intelligence for your Cognigy chatbot “

Our solution is, so to speak, an intelligent search engine (internal Google), rights-checked, scalable, with linguistics and AI for best usability, interacting with your Cognigy chatbot.

Our iFinder system can easily access a wide range of systems such as M365, Confluence, Sharepoint, Jira, Wikis and many more in your company via so-called connectors.

Additionally, an AI knowledge management solution can also enable better management and maintenance of knowledge by providing an easy way to add, update, or remove new knowledge to ensure the chatbot stays current.
Overall, a good AI knowledge management solution can improve a chatbot’s intelligence by giving it a broader knowledge base and a better way to access it to provide appropriate and relevant answers to user questions.

The maintenance effort is reduced because the iFinder already provides many concepts “out of the box”, which reduces maintenance in the “intents”, i.e. questions.
It also draws on existing knowledge from the available information.
For example, the search can add synonyms to the search time, which then, for example, from a search query for pension automatically also searches for pension , old-age pension, company pension, statutory pension, old-age insurance, old-age pension and delivers corresponding hits.
Also, already in the textual preview the context to the searched answer becomes clear.
Relevance Ranking can be used to fine-tune what the best answer is from the available information.

Integrations Components

Does this also work in conjunction with ChatGPT?

A generative language model like ChatGPT cannot replace a search engine or knowledge base because it does not have up-to-date factual and detailed knowledge. Rather, generative language models like GPT4, are the perfect complement to search engines. A search engine can provide passages of text from documents that are likely to provide answers to a posed question, and the language model can use this to formulate an understandable answer that can be used in a chatbot.

Through our flexible integration components, we can also integrate your chatbots and virtual assistants from vendors such as Genesys, google dialogflow, IBM-Watson, moin.AI, Parloa, RASA, Solvemate, OMQ, Userlike, Zendesk and many more.

From there, combining an AI-based search engine with automated text extraction, like iFinder, and a generative language model, like ChatGPT, is the perfect solution for businesses. Here, ChatGPT handles the human-like, automated dialog and iFinder extracts the relevant content from the companies’ data sources.

Search Engine Architecture

The platform is used by customers such as BMW, E.ON, Siemens, Comcast, Anthem, Vodafone, Audi, Bosch, National Underground Group, Daimler, John Hopkins University, Bizerba, RollsRoyce.

Let us discuss your individual use case in person. We are experts in the field of knowledge management, have been dealing with different LLM -Large Language models (besides GPT4 also LaMDA, BARD or MT-NLP etc.) for a long time and provide you with different implementation proposals with advantages and disadvantages, so that you get the optimal decision basis for your project.

Now we are very curious how your current ChatBot application looks like and we are looking forward to discuss further optimization potentials with you and to implement them together.