GenAI for Business Users

What are LLMs

• Learn the concept of tokenizing text, understand context via vectors, and capture data into the weights during training.

The concept of LLM OS

• In the future, you will have an LLM on every device. LLMs will likely be able to carry out tasks in different applications, so it's important to master prompting them.

Addressing limitations

• Learn to handle data cutoff, bias, factual correctness, context size, data privacy, price, and more.

The foundations of effective prompting

• Learn how to provide as much detail and context as possible. Use the prompt ingredients and the techniques to reduce hallucinations.

Basic use cases

• Understand how to apply the Socratic method, the Feynman technique, decision making, brainstorming, first communication drafts.

Group practice

• Brainstorming high impact business tasks and learning to write a slightly more complex prompt as an assignment.

Assignment

• Prompting competition.

Starting a prompt from scratch

• Learn how to state your problem, your goal & let the LLM advise you how to prompt. Understand how to use external prompt libraries and build your own library.

Getting buy-in

• Start using a handy framework for prioritizing. How to build a business case for customer feedback labelling.

Qualitative and quantitative data analysis

• Learn how to generate synthetic datasets. Labelling zero shot. Labelling few shot. Compare results and inform the business case.

Creating a first draft from file

• How to build a weekly email report. Learn to use your core metrics and sample or synthetic data to test and improve the insights that should go into it.

SQL queries

• Learn how to prepare a table structure to generate reliable SQL.

Web scraping competitor reviews

• Learn how to confidently scrape data from a website.

Assignment

• Build a business case & do a POC for your department’s #1 most time-consuming yet feasible task.

Prompt engineering vs RAG vs fine-tuning

• Learn how to instruct the model to perform tasks vs referencing/searching additional data while instructing.

Custom GPTs vs Assistants API

• We'll look at the differences between how a non- technical user would leverage Custom GPTs vs how a developer would do the same via the Assistants API.

Custom GPTs

• First POC battlefield and first scaling platform for your automations.

Building our own custom GPT

• We'll follow the steps, starting from setting up the instructions, through adding knowledge files (RAG), through fixing specific use cases in the prompt and improving its security.

Q&A & Group practice

• Discuss business cases.

3rd-party chatbots

• We'll create a chatbot that we can embed on our website by essentially doing the same as with a custom GPT but through a 3rd party that wraps its solution around the OpenAI Assistants API and allows you to embed the resulting chatbot.

Assignment

• Create your custom GPT.

The SPARK Framework

• Spot potential, Pick appropriate models (Foundation models (ChatGPT, Gemini, …), 3rd-Party Tools, Fine-tuning), Apply and Adapt (Human in the loop), Ready the team (AI policy, Data Privacy, Technical Training), Кeep evaluating progress.
• Examples / Business cases: Ashley, Jemma, Drew, Pheonix, Brad, Tagify, Nina Morisson, Apollo