Make it simple, using generative AI to accelerate learning

What if your team’s tribal knowledge was instantly searchable? Learn to build a custom AI assistant that eliminates outdated docs and accelerates onboarding.

#1about 1 minute

Software development is fundamentally a learning process

Foundational knowledge remains essential for technical professionals even in the age of generative AI.

#2about 2 minutes

Understanding the technical professional's learning loop

The learning cycle of identifying problems, gathering information, studying, and applying knowledge faces challenges like information overload and debugging failures.

#3about 1 minute

Solving onboarding pains with centralized documentation

The challenge of outdated or missing documentation during onboarding can be solved by centralizing information and enabling natural language queries.

#4about 3 minutes

An overview of generative AI and its capabilities

Generative AI is a subset of deep learning that uses large, pre-trained models to perform tasks like text generation, summarization, and creating multimodal content.

#5about 3 minutes

Addressing LLM challenges with model customization techniques

Overcome common LLM issues like hallucinations and knowledge cutoffs by using customization methods ranging from prompt engineering and RAG to fine-tuning.

#6about 1 minute

How retrieval-augmented generation (RAG) works

RAG enhances model responses by ingesting data into a vector store and then retrieving relevant information to augment the prompt at inference time.

#7about 2 minutes

Exploring the AWS generative AI stack for developers

The AWS stack provides tools at every level, from ready-to-use applications like Amazon Q to foundational services like Amazon Bedrock for building custom solutions.

#8about 3 minutes

Demonstrating LLM hallucinations with a basic chatbot

A standard chatbot without proper context or guardrails can hallucinate and provide incorrect information, highlighting the need for human verification.

#9about 4 minutes

Building a custom RAG chatbot with guardrails

Create a more reliable chatbot by defining a system prompt with clear instructions and connecting it to a knowledge base of official documentation.

#10about 3 minutes

Testing the RAG bot for accurate, sourced answers

The custom RAG bot successfully generates accurate code and answers questions by retrieving information from its knowledge base and citing its sources.

#11about 2 minutes

Addressing AI limitations with human oversight and security

Mitigate risks like bad data and privacy concerns by implementing a human-in-the-loop process and leveraging security features like AWS PrivateLink.

#12about 1 minute

Best practices for adopting generative AI responsibly

Successfully integrate generative AI by defining narrow use cases, training teams on proper usage, and prioritizing accuracy and security in all applications.