AI Core Terminology &
Token Management

A practical guide for the team to understand how AI works under the hood and how to make it work efficiently for your projects.

ConceptsEfficiencyBest Practices

The Core Terminology

To master AI tools, you first need to understand the four fundamental concepts that dictate how they process and remember information:

1. Tokens

The fundamental currency and building blocks of AI language.

2. Context Window

The short-term memory limit for a single conversation.

3. Parameters

The "brain cells" determining complexity and intelligence.

4. Memory

Long-term storage for facts across separate chat sessions.

1. Tokens

The Currency of AI

AI models do not read words the way humans do; they break text down into smaller pieces called tokens.

  • What they are: A single character, syllable, word, or part of a word.
  • Conversion: 1 token ≈ 4 characters ≈ 0.75 words.
  • Example: "AI is amazing" = ~3 to 4 tokens.
  • Why you care: Models have strict limits. If you pay for an API, you are billed based on tokens sent (input) and received (output).
01 10 DATA TOKENS

2. Context Window

Short-Term Memory / RAM

Think of the context window as the AI's active, short-term memory during a single conversation.

  • What it is: Maximum tokens the AI can "read" and remember simultaneously.
  • Includes: Your current prompt, past Q&A history, and hidden system instructions.
  • Why you care: If a chat goes on too long or you upload a massive file, you exceed this limit, causing the AI to "forget" earlier context.
ACTIVE MEMORY CONTEXT BUFFER

3. Parameters

The Brain Cells

Parameters are the internal settings and weights that the AI learned during its training phase.

  • What they are: Model labels like 8B or 70B stand for Billions of parameters. They define the model's complex knowledge base.
  • Why you care: More parameters = a smarter, more capable model.
  • The Trade-off: Higher parameters require massive computing power, leading to slower responses or higher costs.
70B WEIGHTS & NODES

4. Memory

Long-Term Storage

While the context window handles a single chat, some advanced AI assistants feature persistent Long-Term Memory.

  • What it is: A background system that extracts and saves key facts about you.
  • Examples: Your name, your job role, your coding environment, or formatting preferences.
  • Benefit: These facts persist across entirely separate conversation sessions, saving you from repeating yourself.
PERSISTENT STORAGE

When the Context Drops

Because AI models have limited short-term memory, they must manage it when a conversation goes on too long. They do this by "forgetting" the oldest parts of the chat to make room for your newest message.

How to Spot a "Lost" Context Window

1. The AI contradicts itself

If you said "No Python" early on, and 50 messages later it outputs Python code, the original rule slid out of the window.

2. Loss of persona

If you requested a "strict code auditor" and it reverts to a generic, polite assistant, it forgot its initial role constraint.

3. Hallucinations or confusion

At the absolute edge of limits, it may repeat itself or generate gibberish because it lacks the conversation history.

4. Explicit UI indicators

Advanced platforms will warn you visually, clear old messages, or stop you from typing when max tokens are hit.

Maximizing Efficiency

To get the most out of your AI sessions, you need to minimize token waste and force conciseness. Stop the model from wasting tokens on assumptions and pleasantries.

Strategy 1:
Wait for Confirmation

Stop the model from giving you steps 2, 3, and 4 before you've tested step 1. Explicitly dictate the workflow with a hard stop.

PROMPT EXAMPLE

"I need to extract journal transaction logs from the iSeries DB. Give me ONLY the first step to establish the connection. Stop generating and wait for me to confirm it was successful before providing the next step."

USER CHECKPOINT

Strategy 2: Divide & Conquer

Break large tasks down into manageable steps and focus on one thing at a time. This prevents overwhelming the AI's context window.
Example: Building a new application.

1

Discussion

Instruct the AI that initially, there is no code. Discuss the high-level goal and outline what needs to be achieved first.

2

UI Design

Collaborate solely on the user interface and user experience design before touching backend logic.

3

Tech Stack

Decide on the specific technologies (database, language, framework, hosting platform) needed for the project.

4

Code & Test

Finally, move on to writing the actual code, making iterative modifications, and conducting tests.

Strategy 3:
Ruthless Formatting Constraints

Models are trained to be polite, leading to padded answers (e.g., "Certainly! Here is your code..."). Strip this out to save tokens.

Effective Constraint Phrases

  • "No yapping."

    Surprisingly effective on newer models to cut out conversational filler.

  • "Output only the SQL query."

    Followed by: "Do not provide explanations, markdown formatting, or introductory text."

  • "Answer in a maximum of two sentences."

    Forces the model to summarize its internal thoughts heavily before outputting.

More to Come ...

"You may think that Randy dedicated a lot of time to put this together to aide the team in AI interactions, but he actually had me do it. My creator Google calls me Gemini, Randy can be sneaky ... gotta watch that guy ..😉"

1 / 13