AI, Amped 101 Cheat Sheet: Terms to Know and Habits That Get Better Results
People hold off on AI because they think they need to understand it first. But the best way to learn is by diving in and actually using it. Oftentimes, the tools themselves can teach you most of what you need to know.
That was the starting point for the first session of AI, Amped, VideoAmp’s webinar series on using AI at work. Kerri Anderson, VP of Sales AI Enablement, and Eric Blenk, Senior Director of Client Enablement, covered the basics: the terms you’ll hear in every meeting, the habits that make AI useful from the first week, and what to check before you trust an answer.
With Kerri’s background in sales and agency planning and buying, not engineering, her advice comes from a place of figuring it out with a hands-on approach.
We pulled the most useful parts into one page you can bookmark and come back to.
In this post, you’ll find:
- Plain-language definitions of seven AI terms you’ll hear often
- Two habits to build when you’re starting out
- How to give AI the context it needs, using a “new hire” mindset
- Three questions to ask before you put your name on AI output
Seven AI terms, in plain English
You don’t need a technical background to follow AI conversations. Most of the jargon maps to ideas you already use at work.
| Term | What it means |
|---|---|
| Large language model aka LLM | The engine behind AI chat tools. It’s trained on huge amounts of text and writes responses by predicting language. Whatever AI assistant you use, an LLM is doing the processing underneath. |
| Prompt | Anything you type or say to an AI tool to get a response. It can be one line or a few paragraphs, and it doesn’t have to be perfect to work. |
| Context | The background you give AI about who you are, what you’re working on and what a good result looks like. More context usually means a better answer on the first try. |
| Agent | AI that completes a multi-step task on its own once you give it the goal. A chatbot tells you what time the movie starts. An agent finds the showing and buys the tickets. |
| Skill | A saved set of instructions that teaches an AI tool to do a specific task your way, like writing in your voice or following a team checklist. You write it once, and the tool uses it every time that task comes up. |
| Hallucination | When AI produces something that sounds confident and polished but is wrong. It’s the main reason to review AI output before you share it. |
| Model Context Protocol aka MCP | An open standard that connects AI tools to other software and data sources. It lets the AI pull context from those systems and take actions in them. |
Two habits to start with
1. Get to know your AI tools, and let them get to know you
The more your AI understands about how you work, what you do and what you’re trying to accomplish, the more useful it gets. That takes some time up front. The next section covers how to do it.
It also helps to know what each tool is good at. You might use one as a thinking partner for strategy and learning, one for building things like visualizations, workflows and websites, and one as an agent that handles errands you’d rather not do, like booking reservations or filing a pet insurance claim. Try a few, notice where each one does well, and match the tool to the task.
2. You don’t need to know everything to get started
The most common blocker for people starting out is “I don’t know enough yet.” Curiosity matters more than expertise here, because the fastest way to learn AI is to use AI.
If you’re stuck on how to do something, ask the tool. If a new term keeps coming up in meetings, ask what it means and how other people are using it.
Use AI to keep up with how fast AI changes. One option is a short daily briefing your AI sends every morning. For example, it could include:
- The three biggest AI headlines from the day before, written in your terms, with a note on whether each one is hype or something worth watching and how it relates to your job
- One breakthrough in your industry that used AI, such as a medical advance if you work in pharma advertising
- One podcast episode or video to watch, chosen by the AI, so you hear from a wider range of perspectives than a single show would give you
You can set up something similar in a few minutes. Tell your AI what you want to track, how you like to receive information and how long it should be.
Treat your AI like a new hire
When people say their AI tool “won’t do what I want,” the cause is almost always the same: the request didn’t include enough context. A message like “I need a spreadsheet on this” leaves the AI guessing about what goes in it, who it’s for and what done looks like.
Think about a new hire on their first day. They know a little about the company and a little about you. Managers who spend time up front explaining the role, sharing resources and setting expectations get a productive teammate faster. AI works the same way.
The same applies when you start with a new work AI tool that has none of your history in it. The first results can be frustrating. Here’s how to get it up to speed:
- Write a career profile. If you already use an AI tool personally, ask it to help you draft a professional-only summary of your role, how you work and what you’re responsible for.
- Load it into your work AI. Most AI tools have a place for standing preferences or instructions, and many let you attach reference documents to a project. Use both, so the context loads once instead of in every chat.
- Create a voice skill. It should spell out your tone, your writing preferences and what you expect from a draft, so anything written for you sounds like you.
- Keep coaching. When output misses, ask the AI why. For example, if you’ve told it not to use em dashes and it keeps using them, asking why can surface unclear wording in the skill. The AI can then rewrite the rule and update the skill.
The results should improve right away, but the work isn’t finished on day one. Like any new hire, your AI gets better with feedback.
Talk to it the way you’d talk to a colleague
Your prompt should describe what you’re trying to do from start to finish, the way you’d brief a coworker. Try using voice input on your phone and just talking it through. Over time, the AI picks up on your tone and how you think.
Turn repeat work into a skill
If you find yourself giving the same instructions over and over, that’s a candidate for a skill. You don’t need documentation to build one. You can build your first skill by asking AI how skills work and how to make one.
If you’ve been using AI for a while, try this prompt:
“Based on what I’ve been working on recently, is there a repeatable workflow I should turn into a skill?”
Check the work before you put your name on it
A confident, polished answer from AI is not the same as an accurate one. AI can hallucinate, producing something that reads as correct but isn’t. That makes human review part of the job, not an extra step.
Try a simple test. Imagine a client comes back and tells you something in your deliverable is wrong. Your name is on it. Are you comfortable with that?
Before you share AI output, ask:
Where did this information come from?
Know whether the AI pulled from your files, a connected system, the web or its own training.
Did the AI have the data it needed?
If it didn’t, it may fill the gap with something plausible.
Would I stand behind this if someone challenged it?
If the answer is no, check the sources or rewrite it.
This matters a lot in advertising. You can ask AI to analyze how a campaign performed, but if it doesn’t have the right measurement data, or doesn’t understand how a metric is defined, the analysis won’t be worth much. Asking AI good questions is only half the work; the other half is knowing where the data behind the answer comes from.
It’s also worth remembering that everyone is at a different stage. Across agencies and advertisers in the U.S., some teams are just starting to try the AI tool their company provides, while others run data science groups that build advanced workflows and agents. There’s no single right approach, and you don’t need everything figured out to start.
Keep learning with AI, Amped
This was the 101. Upcoming AI, Amped sessions will go deeper on agents, MCP and more advanced ways to put AI to work.
Our measurement solution is designed to give you numbers you can trust, and that’s the base for the AI features in our platform too. In the session, Eric showed how VideoAmp customers can open an outcomes report and ask questions in plain language: summarize the report, generate new visualizations or pick up a past conversation, with data export still available for anyone who wants to work with the numbers directly.
Want the full session? Request access to the AI, Amped recording to watch Kerri and Eric walk through it in 34 minutes, including Eric’s demo of AI in VideoAmp reporting.