Beginner · Level 1: Foundations
What Is AI? A Plain English Guide for Salespeople
For a salesperson, AI usually means an AI assistant: a writing and thinking tool that learned from a huge amount of text and produces the most likely good answer to what you ask. Think of it as a brilliant new hire on day one: it knows a lot about the world and nothing about your company until you tell it.
In 30 seconds
- AI assistants such as ChatGPT, Claude, Gemini and Microsoft Copilot are writing and thinking tools that learned from a huge amount of text.
- Picture a brilliant new hire on day one. It knows a lot about the world, and nothing about your company, buyers or pricing until you tell it.
- What it learned in training is frozen in time. What you give it today is called context, and context decides how good the answer is.
- It's strong at language and structure: drafting, summarizing, organizing and practice. It's weak on facts nobody gave it.
- Like any new hire's work, you check it before a buyer sees it. AI wrote it; you're responsible for it.
AI in one sentence
When salespeople say "AI" today, they almost always mean an AI assistant: a chat tool you type or talk to, built on a large language model. That model learned patterns from an enormous amount of text, so it can write, summarize, explain and reason in plain language.
It isn't a search engine or a database of facts about your accounts. It produces the most likely good answer from what it learned and what you give it.
Meet your new hire
Picture the smartest new hire your team has ever had, on their first morning. They read fast, write well and know a little about almost everything. They have never heard of your product.
Scroll sideways to see the whole table.
| A brilliant new hire on day one... | So with AI, you... |
|---|---|
| Knows a lot about business, industries and how sales works in general | Can ask it to explain a buyer's industry or a term on an earnings call |
| Writes clean drafts quickly and never minds a rewrite | Can ask for three versions, then ask for a fourth without guilt |
| Has never seen your product, pricing, best fit customers or best emails | Have to tell it, or it fills the gaps with generic guesses |
| Missed everything that happened before they started reading the news | Can't trust it on last week's funding round unless it can search the web or you paste the source |
| Wants to look helpful, so rarely says "I don't know" | Must check every fact before a buyer sees it |
Key idea
AI is a brilliant new hire on day one. It knows the world, but not your business, until you tell it.
Training versus context
A new hire brings two kinds of knowledge: what they learned before they joined, and the onboarding binder you hand them today. AI works the same way.
- Training is what the model learned before you ever opened it. It's broad, but it stops at a date called the knowledge cutoff. Anything newer is a blank unless the tool can search the web.
- Context is everything the model can see right now: your request, files you upload, earlier messages in the chat and any tools connected to it. The amount it can hold at once is called its context window. It's large, but not unlimited.
When an answer comes back bland or wrong, the model usually isn't the problem. The context is. You left a gap, and AI filled it with the most generic plausible guess. Say a buyer at Harlow Medical Supply, a fictional company, has just told you "we already have an onboarding program."
Before: no binder
A prospect said "we already have an onboarding program." How should I respond?
What comes back: a generic objection script that could fit any product and any buyer. It may even promise results you can't back up, because it has nothing real to work from.
After: the binder is open
I'm a closing rep at Ramplight (a fictional company). We help sales teams ramp new reps faster with AI coaching. On today's call, the sales director at Harlow Medical Supply (also fictional) said "we already have an onboarding program." My call notes are below. Using only what's in my notes, suggest 3 questions that help me understand what their program does well and where new reps still get stuck. Don't argue with the buyer.
[paste your call notes]
What comes back: questions built on what the director actually said, instead of a canned script that argues with the buyer. Same tool, same minute. The only difference was the binder.
How the model turns training and context into an answer, and why it gets things confidently wrong, is the next page on this path: How AI Assistants Actually Work. To write a full request like the "after" prompt every time, see Prompting 101.
What to hand AI, and what to keep
You wouldn't give a day one hire your biggest renewal, and you wouldn't leave them idle either. You'd hand them work that suits them, check what matters and keep the judgment calls. Sort AI the same way.
Scroll sideways to see the whole table.
| Give it freely | Give it, then check every fact | Keep it yourself |
|---|---|---|
| Summarize a long earnings call or a 60 page annual report you pasted in | A prospect email or LinkedIn message | Pricing, discounts and contract terms |
| Draft ten subject lines or three angles for an opener | An account brief before you use it on a call | Security, legal or compliance answers |
| Play a skeptical buyer so you can practice objections | A recap email after a customer call | Reading the room in a live conversation |
| Turn messy call notes into a clean structure | CRM notes that other people will rely on | Deciding which deal deserves your Friday afternoon |
The pattern is simple. AI is excellent with language and structure. It's unreliable on facts it wasn't given and has no real feel for people, politics and timing inside a deal. Those stay with you. If you've heard that AI can do far more (or far less) than this, 10 Myths About AI in Sales sorts the claims from the facts. For an honest look at the limits, see What AI Can't Do in Sales (Yet).
Try the same sort on your own week in your company approved AI assistant. It takes five minutes and shows you where AI fits first.
Prompt
Sort your week
Here are 10 tasks from my week as a [prospecting or closing rep] at [your company]. Sort each into one of three groups: give to AI freely, give to AI with a fact check, or keep myself. Give a one line reason for each.
[list your 10 tasks]Where the new hire picture breaks
Every analogy has limits. Knowing these four keeps you out of trouble.
- It can't tell what it invented unless you ask. How AI Assistants Actually Work shows a quick way to trace where each fact came from.
- It may forget you tomorrow. A new hire remembers last week's coaching. In many tools, each new chat starts fresh unless you save your instructions in a project, custom instructions or a memory feature. Check how your tool works.
- It's faster than any human. It reads a long document in a minute and never tires of rewrites, so ask for options instead of settling.
- It has no stake in the deal. It won't lose sleep over a missed number or a burned relationship. You will, which is exactly why the final call is yours.
How much to trust it, and when
A good new hire earns responsibility over time. AI should too.
Assist
AI drafts, summarizes and answers. You copy, edit and send. Example: "Write a first line for this email."
Copilot
AI uses connected tools and proposes actions. You approve each one. Example: "Update these five CRM fields from my call notes."
Agent
AI runs multistep work on its own, often on a schedule. You review what it did. Example: "Every Monday, brief me on deals that haven't moved."
Start at step one and move up only as fast as your checking habits do. At every step, treat AI's work like a new hire's first proposal: nothing reaches a buyer until you've read it, fixed it and put your name on it.
Quick check
Quick check
Answer in your head, then open each one.
You ask AI about a prospect's funding round from last week, and it gives you a specific amount. What should you assume?
Answer: That it may be a guess. Training stops at a cutoff date, so unless the tool searched the web and you opened the source yourself, treat the number as unverified.
Which of these is training and which is context: AI knowing what a VP of Sales usually cares about, or AI using the call notes you just pasted?
Answer: Knowing what a VP of Sales usually cares about is training: broad, general and dated. Your pasted call notes are context: specific, current and the reason the answer fits your deal.
Which of these should you keep for yourself: summarizing a webinar transcript, drafting subject lines, or answering a prospect's question about contract terms?
Answer: Contract terms. That's a high stakes answer that needs approval from the right owner, not an AI draft.
FAQ
What is AI, in simple terms, for a salesperson?
AI is a writing and thinking assistant that learned from a huge amount of text. For a salesperson, think of it as a brilliant new hire on day one: great at drafting, summarizing and practice, but it knows nothing about your company or buyers until you tell it.
Is ChatGPT the same thing as AI?
ChatGPT is one AI assistant among several. Claude, Gemini and Microsoft Copilot work in a similar way, and the habits on this page apply to all of them.
Does AI know about my company and my accounts?
Not unless you tell it or connect it to your tools. Its training covers general knowledge, so details about your product, pricing and prospects have to come from you as context.
Why does the same AI tool give one rep great answers and another rep generic ones?
The difference is almost always context. The rep who gets sharp answers tells AI who they are, who the buyer is and what material to use. The other leaves gaps, and AI fills them with generic guesses.
What is AI best at in sales?
AI is best at language and structure: summarizing long material, drafting emails and call plans, generating options and playing a buyer so you can practice. It's weakest on facts nobody gave it and on judgment calls inside a live deal.
AI for Sales Certification
Turn your new hire into your best teammate
10 self paced sessions, an optional weekly live class and a graded capstone, all practiced on your own accounts. Membership is $50 a month. Add the $250 certification when you're ready.
Not sure where to start? Take the 2 minute assessment