# AI Tool Foundations

URL: https://www.onsomble.ai/guide/foundations

Updated: 2026-09-14

Build the mental models that change how you use every AI tool.

<div>
<div>For anyone who uses AI tools but doesn't really understand what's happening underneath. The goal is correct mental models that change how you use AI, not academic definitions.</div>
<br>
<div>After completing Tier 1, you will understand what's actually happening when you use an AI tool, know how to choose the right tool for a task, prompt effectively, and evaluate whether AI output is trustworthy &mdash; with a clear framework for each.</div>
</div>

Level: Beginner · 3.1666666666666665 hours

## You're Already Using AI. Here's What's Actually Happening.

You've used ChatGPT, Claude, Copilot. But do you know what actually happens when you type a prompt? This module builds the mental model that changes everything.

- Distinguish between AI, machine learning, and generative AI in practical terms
- Identify which type of AI powers the tools you already use
- Explain to a colleague why AI is not one thing, and why that matters for how you trust it

[Read this module](https://www.onsomble.ai/guide/foundations/youre-already-using-ai)

## How AI Learns (Without Anyone Programming the Answers)

Traditional software follows rules. Machine learning discovers them. That shift explains most of how modern AI behaves, its strengths and its failure modes.

- Explain how machine learning differs from traditional programming
- Understand why training data quality directly affects AI output
- Recognise where bias enters AI systems and why it is structural, not accidental

[Read this module](https://www.onsomble.ai/guide/foundations/how-ai-learns)

## How LLMs Work: The Engine Under Every AI Tool You Use

An LLM does one thing: predict the next word. That single idea explains hallucination, creativity, prompt sensitivity, and every other behaviour that surprises AI users.

- Explain next-token prediction and why it matters for how you use AI
- Understand what tokens and context windows are and why they affect your results
- Recognise why LLMs are confidently wrong, and why that is a feature of the architecture, not a fixable bug

[Read this module](https://www.onsomble.ai/guide/foundations/how-llms-work)

## The AI Tool Landscape: Models, Apps, and How to Choose

The AI landscape is confusing because the same word describes fundamentally different products. This module gives you a framework for choosing the right tool, understanding what you're paying for, and knowing what happens to your data.

- Distinguish between models, platforms, and applications, and explain why it matters
- Evaluate which AI tool fits a given task using a practical decision matrix
- Understand what you gain and lose with free vs paid tiers, and what happens to your data

[Read this module](https://www.onsomble.ai/guide/foundations/ai-tool-landscape)

## Prompt Engineering: The Skill That Changes Everything

Most people prompt AI like a search engine. This module teaches you to prompt it like a collaborator, with a framework that works across every tool.

- Explain why the context you provide matters more than how you phrase the question
- Build effective prompts using a five-element framework you can apply to any AI tool
- Choose the right prompting technique for a given task (and know when simpler is better)
- Diagnose why a prompt failed and fix it without starting from scratch

[Read this module](https://www.onsomble.ai/guide/foundations/prompt-engineering)

## RAG, Agents, and How AI Connects to Your World

AI stopped being a text box. RAG connects it to your documents. Agents let it take action. This module explains the shift from answering to doing.

- Explain what RAG does and why it produces better answers than a base LLM alone
- Describe the three-step RAG pipeline (chunk, retrieve, generate) and where it breaks down
- Distinguish between AI that answers questions and AI agents that take action
- Evaluate when a tool needs RAG, an agent, or just a bigger context window

[Read this module](https://www.onsomble.ai/guide/foundations/rag-agents-tool-use)

## Evaluating AI Output: A Trust Framework for Knowledge Workers

Not all AI output needs the same level of scrutiny. A practical framework for knowing what to verify, what to spot-check, and what to trust.

- Apply a three-tier verification triage to any AI output (always verify, spot-check, or generally trust)
- Identify when AI is most and least likely to be reliable, based on task type and domain
- Spot the three most common failure modes: hallucination, bias, and shallow reasoning
- Build a personal verification workflow you can use at work this week

[Read this module](https://www.onsomble.ai/guide/foundations/evaluating-ai-output)

## Responsible Use: Ethics, Data, and the Business of AI

Your AI inputs go somewhere. Your outputs may not be yours. This module covers data privacy, copyright, workplace policies, AI economics, and building a personal responsible use framework.

- Explain what happens to your data when you use free vs paid AI tools
- Describe the current state of copyright law for AI-generated content
- Evaluate whether your organisation's AI policy covers the risks that matter
- Build a personal framework for deciding what to delegate, what to keep, and how to stay sharp

[Read this module](https://www.onsomble.ai/guide/foundations/responsible-ai)
