Adis: creating your company's knowledge base and exploiting it through deepseek

One of the main objectives of artificial intelligence (AI) in the current context, marked by the high interest in language models (LLM), is to become a useful tool for companies. The concept of a knowledge base and business wisdom is not new; it has been developed through solutions such as intranets, teams, SharePoint, SQL and NoSQL databases, media, big data, among others.

There is currently a significant opportunity to combine all this information and leverage it through LLMs and generative AI (GEN AI). This approach represents a competitive advantage by having this information trained and ready to be used. The first steps include automated integration of this information and comparing it to market trends and customer needs.

Although these topics have been discussed for years in relation to business intelligence, big data, and machine learning strategies, the situation is different now. The computing power of GPUs allows neural networks to be maximized, providing tangible benefits.

Steps include:

1. Connect the entire universe of company information through intelligent automation of pipelines, web scraping, APIs, etc. This task is laborious and delicate, but AI's predictive models and architecture can make the process easier.

2. Feed the universal business knowledge base using, for example, a base of graphs and deep learning models to evaluate and qualify actions. You can use in-memory databases for faster query speed or solid-state disks, and do everything in Python to keep costs down.

3. Connect and train this model and knowledge base with the most important LLMs, whether on-premise or in the cloud, ensuring high levels of security and monitoring to avoid information leaks.

   
   

Benefits:

• GEN AI for human resources and process training.

• Suggestions for optimization and qualification of operational processes.

• Anti-fraud monitoring.

• Automatic corporate reports and generation of dashboards.

• GEN AI for CRM.

• AI agents for operational processes connected to workflows.

• KPIs virtual assistant.

• Internal knowledge management.

• Intelligent assistance with business data through messaging.

• Qualification and feedback in decision making.

• Personalized sales.

• Digital twins in factories.

  
  

These are just some of the benefits, although there are many more. The first step to achieving an information-oriented enterprise exploitable by AI is the creation of a universal base of knowledge and business wisdom. The sale of information can become a new product for the company, so it is important to manage a viable architecture and keep costs under control.

For technological infrastructure, various options can be considered, from open-source solutions such as Python, DeepSeek, and PostgreSQL, to well-known services such as Azure, AWS, GCP, or a combination of these. The role of the architect is crucial to dimension these solutions and present economic viability.

From a strategic perspective, it's not a matter of if companies will do it, but when. This requires a lot of human expertise, as well as architects and directors who generate a blueprint and a methodology for its implementation. AI and its applications offer new opportunities in this area.