In 2025 What Should You Learn In AI ?

Summary

In this video, Krishna explores the 2025 AI Engineering Report, which highlights current trends in generative and agentic AI. The report emphasizes the growing importance of Retrieval Augmented Generation (RAG) applications, the integration of human-in-the-loop systems, and the rising adoption of audio and fine-tuning methods for AI models. Krishna discusses the top AI frameworks like LangChain and Llama Index, stressing the need for professionals to stay updated with AI advancements. The video serves as a guide for anyone aiming to build a career in AI, particularly in developing RAG-based applications and leveraging cutting-edge AI technologies.

Takeaways

• 😀 Generative and agentic AI are highly popular topics right now, with advancements in LLM models and cost reduction, especially with GPT-5 from OpenAI.
• 😀 The 2025 AI Engineering Report provides valuable insights into where to focus for a career in AI, offering guidance for those aiming to make a successful career transition.
• 😀 The report surveyed hundreds of engineers working in AI, providing detailed information on models, workflows, and AI use cases.
• 😀 OpenAI's models dominate the market for customer-facing applications, followed by Anthropic's Cloudy model, especially for RAG (retrieval-augmented generation) applications.
• 😀 RAG (retrieval-augmented generation) is a key area of focus in the AI field, with many companies leveraging it for automation and building agentic applications.
• 😀 More than 70% of the respondents are using RAG in some form, and there are various types, such as autonomous and adaptive RAG.
• 😀 AI model updates are frequent, with over 50% of respondents updating their models at least monthly, often including fine-tuning to improve performance.
• 😀 Audio integration in AI workflows is on the rise, with 37% of respondents planning to adopt it soon.
• 😀 The concept of 'human in the loop' is widely used in AI applications, particularly in agentic AI models, allowing human input for decision-making.
• 😀 Top AI use cases include code intelligence, content generation, customer support, fraud detection, and workflow automation, all of which are critical in AI's application in industry.
• 😀 Langchain and Langraph are key frameworks for building AI applications, with a focus on monitoring, observability, and security, making them essential tools for AI developers.

Q & A

What is the main focus of the 2025 AI Engineering Report discussed in the video?

The main focus of the 2025 AI Engineering Report is to provide insights into the current trends in AI engineering, including the most popular models being used, the growing importance of generative AI and agentic AI, and the specific areas where professionals should focus to advance in AI careers.

What is the significance of RAG (Retrieval-Augmented Generation) in the AI industry according to the video?

RAG is significant because it is increasingly used to automate workflows and create agent-based applications. It is a primary focus for many companies and has various applications, including autonomous, self, and adaptive RAG models. The video emphasizes the importance of learning RAG as it is a major trend in AI development.

What percentage of respondents are using RAG in some form?

70% of the respondents reported using RAG in some form, highlighting its widespread adoption in AI-driven applications.

How often are most organizations updating their AI models?

Over 50% of the respondents are updating their AI models at least monthly, with some updating their models even more frequently.

Which companies are reported to be using OpenAI models extensively in their applications?

Top companies such as EY, PWC, and KPMG are using OpenAI models for building AI applications, specifically for customer-facing and internal use cases.

What role does 'human in the loop' play in AI systems according to the video?

'Human in the loop' refers to integrating human judgment into the AI process. It is emphasized that most agents in production have some form of human oversight, which helps improve accuracy and decision-making.

What are the most popular use cases for AI applications mentioned in the report?

The most popular use cases for AI applications include code intelligence, writing assistants, content generation, text summarization, workflow and app automation, customer support, fraud detection, and sentiment analysis.

What are the top AI models mentioned for use in production environments?

The top AI models mentioned for use in production environments are OpenAI's models, particularly for customer-facing applications, as well as Anthropic's models, which are also gaining traction.

What are the key frameworks recommended for building AI applications?

The key frameworks recommended for building AI applications include Langchain, Langraph, and Llama Index, which are widely used for developing RAG applications and other AI-driven workflows.

What does the speaker suggest about integrating generative and agentic AI into day-to-day work?

The speaker encourages everyone, from managers to developers, to start integrating generative and agentic AI into their daily work processes. He highlights the vast potential of these technologies and the need to stay updated with AI advancements.