# Ten AI terms you will hear in your first week

Original 5HO practice material. Fictional example. Reviewed 5 October 2026.

## Useful result

You'll finish with a compact reference that points to the full glossary.

You only need enough vocabulary to make the next decision. Nobody is handing out points for saying “orchestration” before breakfast. Use these ten terms to label a simple task.

## Set up

Allow 10 minutes. Read the terms, then paste the example into any AI chat. You do not need to activate all ten capabilities.

## Practice input

Task: use a saved set of instructions to summarise one connected project note, show the source and wait before sharing. Available terms: model, prompt, context, token, memory, project, skill, connector, MCP, agent.

## Do the task

1. Match each term to the definitions in the guide.
2. Describe which capabilities the sample task needs.
3. Remove anything that does not help with this one note.

## Prompt

Explain the sample task using no more than four of the ten terms. Identify the information source, reusable procedure and approval point. Say which features are optional.

## Refine

Do not describe a connector as memory, or a project as an autonomous agent. Give one sentence showing the distinction.

## Expected checks

The source connection and saved procedure have different jobs. MCP may sit underneath a connection, but a learner does not have to configure it whenever they use an app.

## Save

Save a short glossary in your own words. Keep only the terms that help you ask a better question.

## Ten terms, in plain English

- Model: the system generating the response.
- Prompt: the instruction you give it.
- Context: information available for this response.
- Token: a chunk of text or other data used by the model.
- Memory: product features that retain selected information across interactions.
- Project: a product workspace grouping related material.
- Skill: a packaged procedure with instructions and optional supporting files.
- Connector: an integration giving an app access to another service.
- MCP: a standard for connecting AI clients to tools and information.
- Agent: a system that can select and carry out steps with available tools.

## Sources

- Model Context Protocol overview: https://modelcontextprotocol.io/docs/2026-07-28/getting-started/intro