Ourbit Master AI — 15-Day Programme

Ourbit Master AI — Day 5: Context and Memory

Download lesson PDF

Day 5 — Context and Memory

Purpose: Understand context windows, their limits, and practical ways to keep important information available.

Source/build status: Recovered Module 3 is the source; preserve the scrolling-window model.

Learning outcomes

• Explain what a context window makes available to a model during a task.
• Distinguish current context from durable stored or retrieved information.
• Keep critical constraints visible during long-running work.
• Choose when information should be restated, stored or retrieved.

Core lesson — Context and Memory

A language model can only use information made available to the current generation process. A context window is the bounded amount of material a system can consider at once. Context is not the same thing as durable human-like memory. In long work, keep critical constraints explicit and maintain a compact source of truth rather than assuming an old message remains influential forever.

Worked example

A project begins with “Use only free tools.” Much later, a proposed solution includes a paid service. Keep that rule in the current project brief or restate it at the decision point rather than relying on conversational history alone.

Teaching / video direction

Show a window sliding across a long conversation while a compact project brief remains available. Distinguish current context, persistent application state and retrieval. Close by asking how a system finds the right information without keeping everything in view—Day 6.

Hands-on practical exercise

1. Write three non-negotiable constraints for a long AI-assisted task.
2. Create a compact current brief containing those constraints.
3. Continue the task through several exchanges, then deliberately check whether the constraints are still explicit.
4. Identify information that should be stored durably or retrieved only when relevant.
5. Before a consequential decision, re-reference the current source of truth.
6. Record one failure that could occur if old conversational context were assumed to be permanent.

Knowledge check

1. What is a context window in practical terms?
2. Why is context not the same as durable human-like memory?
3. How can a critical constraint be protected during long work?
4. When is retrieval preferable to repeatedly placing everything in context?
5. What can happen when an early instruction is no longer usable by the system?
6. How does Day 5 prepare for semantic retrieval on Day 6?

Assessment and remediation contract

The scored daily quiz uses 10 questions; 7/10 (70%) is a pass. Missed concepts map to exact lesson anchors for targeted remediation, including on a passing attempt. Attempts 1–9 below threshold return to remediation and a fresh eligible question set; an unsuccessful tenth attempt restarts the day in a new learning cycle while preserving history. The 50-question validated bank is a learner-release gate and will be scaled only after the Day 1 runtime passes acceptance.

Production / QA notes

• Use the worked scenario as the primary visual sequence; avoid generic AI-brain imagery.
• Keep terminology vendor-neutral and define unavoidable technical terms in context.
• Do not equate fluent output with verified truth or AI assistance with autonomous authority.
• Before learner release, fact-check the final lesson and validate every scored item against explicit teaching.