How companies can use generative AI – Enterprise Search as an example
What technical challenges await companies looking to adopt generative AI.
Companies are asking themselves how they can use generative AI. But there are many challenges between the idea and the introduction. This blog article shows that with the right systems as a basis, these are not so difficult to master.
Generative Artificial Intelligence (AI) automates the creation of diverse forms of content, from text and images to melodies and designs. It is a transformative technology that represents a differentiated approach for companies to increase efficiency and drive new ways of thinking. Currently, the most relevant for companies is certainly the form of text-generating AI, which is why we will take care of exactly this part in this post.
Index
State of the art
In the first half of 2023 in particular, new AI models were launched on the market practically every week. A wide variety of vendors introduced a variety of products for businesses and consumers in a short period of time. Among them were some sought-after and costly models, featuring specialized functions and cutting-edge technologies. In parallel, many open-source models were released, providing broader applicability and accessibility and allowing companies to customize them to their specific needs. The tech industry is currently still searching for the most efficient model for each use case.
Furthermore, it is likely that models will continue to evolve at a correspondingly rapid pace and applications will need to be built variably for new models accordingly.
Use cases for Generative AI
This blog post is more about the technical hurdles of adoption and less about the potential use cases. Nevertheless, a small overview of possible use cases is given below:
- Product Development / R&D: Generative AI can be used to help engineers understand key information and best practices from projects they’ve already completed, building on existing know-how rather than having to re-engineer it.
- Marketing & Content Creation: generative AI can be used to easily create marketing materials such as blog posts, social media posts, or email campaigns specific to the needs of the target audience.
- Sales: In sales, generative AI can be used to address objections from pot. Customers quickly and easily or to create follow up emails from meeting notes.
- Service (chatbots, self-service): In order to reduce the necessary staff, especially in times of staff shortage, generative ChatBots can be used to answer customer’s questions with the help of internal know-how.
- Human resources: Generative AI can analyze large volumes of resumes according to specific criteria to create a shortlist for the recruiter.
- Enterprise Search: Generative AI can also be used for enterprise searches. In this case, all documents can be indexed and the AI tries to answer answers from employees best possible based on the available internal data.
How can generative systems be connected to internal systems?
The biggest question that arises when implementing a generative system internally is how to combine the internal data with the AI models, taking into account the GDPR.
The following challenges arise:
- Access control: consideration of existing access rights
- Administrative effort: Do documents have to be actively uploaded somewhere or do the systems talk permanently via an interface?
- Hosting: Where should the system run, or does the company have sufficient resources and know-how to administer and maintain such a system?
Most people initially think that AI models need to be re-trained for the internal use case – however, this is not the case. If it were, the following questions would arise:
- How often does the AI model need to be re-trained to always have up-to-date know-how?
- What resources are needed for training (spoiler: many) and who pays for it?
- What data is used for training? Depending on the position/department, the data basis would have to be different.
Since these are not trivial or can be answered with a good ROI, the following solution is a better alternative:
Enterprise Search systems are systems that are already designed to combine information from various internal company systems and display the most relevant information to the employee. One of the standard requirements for an enterprise search is, of course, that access rights are taken into account. This ensures that each employee only finds the information that he or she is actually allowed to see.
By the way: If you are about to introduce an enterprise search, then you are first faced with a make or buy decision. Due to its complexity, enterprise search is usually a buy decision.
Example: Enterprise Search with generative AI
Once the user has given a question or prompt to the system, the enterprise search engine picks out the most relevant information. This is then given to a general (!) generating model with the task of finding the most relevant information, summarizing it and answering the question or prompt in the best way. This eliminates tedious searching and employees quickly and easily get the information from thousands of documents prepared in a relevant way.
Which generating model to choose?
When choosing a generating model, the first question is what happens to the data in the model. If one uses one of the big providers like OpenAI, then the data is processed on their systems. If you use an open source model, you can also host it on your own instances – if you have enough resources. If not, you can often fall back on providers that offer GDPR-compliant hosting of such models. Which model is chosen in an individual case, however, depends heavily on the use case. However, it is important that the architecture is developed in such a way that models can be quickly replaced when more suitable models come onto the market.
Conclusion
The biggest challenge when implementing a generating system is how to connect the internal data to the generating system with little administrative effort, but in a DSGVO compliant way. Enterprise search solutions offer a very good approach here to apply generative AI directly across multiple systems.
Comments are closed.