Sell smarter with AI, and stay safe.
A plain-English guide for sales reps who have never used AI at work. Learn what it is, try safe first steps, and know exactly what to keep out of any chat box.
Your first week, in three steps
You can do this
Nobody was born knowing how to use AI. The reps who get good at it are the ones who try one small, safe thing and keep going. AI handles the busywork so you can spend more time on the part only you can do: listening, building trust, and helping a customer decide.
The traffic-light rule
Every use case in this guide gets one of three lights.
AI basics in five minutes
No technical background needed.
What it is
An AI assistant is a program you chat with. You type a request, called a prompt, and it writes a reply. It predicts likely words from patterns it learned, so it sounds confident whether or not it is right.
What it is good at
Drafting, rewriting, summarizing, brainstorming, practicing conversations, and organizing messy notes.
What it is bad at
Exact facts and numbers, math, live traffic or news, your private deal details, and knowing when it is wrong.
Four rules to start with
- Ask Legal and IT first. They decide which tools and data are allowed.
- Do not paste what you would not post publicly unless the tool is approved for that kind of data.
- You are the author. Check every fact, name, number, and promise before it leaves your hands.
- Start small. Begin with green use cases and move up only with approval.
How to write a better prompt
Give it a role, a task, context, and the format you want.
Write me an email.
Act as a sales coach. Rewrite the follow-up email below for a busy operations manager. Keep it under 90 words and end with one clear question.
Four first prompts to try
Replace the bracketed parts. Use only content that is public or your own, in a tool your company approved.
Words you will hear
- Prompt
- The instruction you type to the AI.
- Model
- The AI engine behind a tool. Claude, GPT, Gemini, Grok, Nova, and Llama are model families.
- Hallucination
- When the AI states something false in a confident tone, such as a fake quote or statistic.
- Business account
- A company-managed plan with a contract and admin controls. Safer than a personal account for work.
- Training
- When a vendor uses what you type to improve its AI. Business plans usually say they do not. Always check.
- Personal data
- Anything that identifies a person, such as names, emails, phone numbers, or health details.
- Shadow IT
- Using tools your company has not approved. The biggest AI risk for sales teams.
AI is not replacing the part of selling you love. It is taking away the part you don't.
Trust, curiosity, and judgment are what close deals, and they are yours. AI is a tool that gives you more time and more room to use them.
It gives time back
Call notes, CRM updates, account research, first drafts, and scheduling add up to hours each week. AI can take a first pass at all of them.
It lowers the blank-page fear
Staring at an empty email is hard. Starting from a draft you can reshape is easy. You stay the author; you just stop starting from zero.
It helps you walk in prepared
Role-play a tough buyer, learn an unfamiliar industry in minutes, and arrive at each call with sharper questions.
Why this is the path forward
- Buyers are already using it. More buyers research vendors with AI before they ever talk to you. Meeting them there is part of the job.
- The routine work is shrinking. What stays valuable is judgment, relationships, and creativity, which are the skills sellers are best at.
- Skills compound. A rep who learns to prompt, check, and refine today gets better every month. The gap between early learners and late ones widens.
- It rewards good habits. Teams that pair AI speed with human review and clear rules get the benefits without the headlines.
How it frees you up
- Typing up call notes
- Hunting for account background
- First drafts of routine emails
- Formatting proposals and recaps
- Cleaning and updating records
- Real conversations with buyers
- Deeper listening and discovery
- Multi-threading into the account
- Following up while interest is warm
- Learning your craft
How outreach becomes scalable
Before AI, personal outreach meant a rep could only research and write so many messages a day. With AI doing the first pass, one rep can give more people a message that is actually relevant to them.
Scale quality, not just volume. Messages that are generic, inaccurate, or unwanted hurt your name and your company's, and may break the law. Keep a human in the loop and honor opt-outs.
Start this week
Pick one green use case, try it once, and note what you learned. After that, check with Legal and IT about which tools you can use for real work.
Three kinds of AI help
Most sales AI falls into one of three groups. Knowing which one you are using tells you how much to check and who needs to approve it. A good path is to start with augmentation, add analysis, and automate last.
AI works with you
You stay in the driver's seat. AI drafts, brainstorms, explains, and coaches, and you decide what to use.
Examples: rewriting an email, role-playing a buyer, preparing call questions, outlining a proposal.
Tools: chat assistants from any of the six vendors, plus assistants built into email, documents, and meetings.
Main risk: trusting a confident but wrong answer. Your job: review every output.
Start level: easiest, mostly green.
AI finds the pattern
AI reads lots of data or text and surfaces what matters, so you can decide where to spend time.
Examples: forecast commentary, lead scoring, call themes, win/loss patterns, churn risk, account prioritization.
Tools: analytics inside your CRM, call analytics tools, spreadsheets and BI tools with AI, and long-document assistants.
Main risk: poor data, bias, and false precision. Your job: sanity-check findings and decide.
Start level: moderate, often amber.
AI does the task
AI completes steps on its own once set up: moving data, routing leads, sending sequences, transcribing calls.
Examples: logging calls, updating CRM fields, routing inbound leads, scheduling follow-ups, running outreach sequences.
Tools: workflow and sales-engagement platforms, CRM automations, and custom agents built by IT on cloud or in-house models.
Main risk: one mistake repeats at scale. Your job: set limits, audit samples, and keep an off switch.
Start level: highest, needs Legal and IT.
| Augmentation | Analysis | Automation | |
|---|---|---|---|
| Who is in control | You, every time | You decide from its findings | The system, within limits you set |
| Biggest benefit | Quality and speed | Focus and insight | Time and consistency |
| Biggest danger | Over-trust | Bad data | Errors at scale |
| Review needed | Each output | Findings and inputs | Samples, logs, and permissions |
| Who sets it up | You, with an approved tool | RevOps or analytics, with IT | IT or RevOps, with Legal |
| Vendor fit (directional) | Chat apps from Claude, GPT, Gemini, Grok | Long-document and data-capable models; Gemini inside Google data tools | APIs: Nova on Bedrock, Llama self-hosted, and Claude, GPT, or Gemini APIs |
AI for go-to-market teams
Go-to-market (GTM) is everyone who finds, wins, and grows customers: sales, marketing, customer success, sales engineering, enablement, partnerships, and revenue operations. Pick a role to see where AI fits and what to watch for.
Where AI fits across the customer journey
The GTM AI tool stack
Most teams end up with several layers. Each has a different owner and risk.
Crawl, walk, run
Who owns what
How to know it is working
A 30-60-90 day plan
Estimate the time you get back
Change the numbers to match your team. The starting values are examples, not benchmarks.
Assumes 46 working weeks. Time back only creates value if it goes into selling activity, and tool costs and training time are not included.
Best practices
Habits that make AI useful and keep you out of trouble.
Ask better
- Give it a role, a task, context, and a format.
- Say who the reader is and what you want them to do.
- Paste a sample of your own writing so it matches your tone.
- Ask for three options, then choose.
- Tell it what to avoid, such as jargon or hype.
- Ask it to ask you questions before it answers.
- Iterate. The second prompt is usually better than the first.
Check everything
- Verify every fact, number, name, and date against a source.
- Ask for sources, then open them.
- Recalculate math and ROI in a spreadsheet.
- Read the whole draft aloud before sending.
- Confirm product, pricing, and compliance claims with the owner.
- Treat confident tone as a warning sign, not proof.
Protect data
- Use only company-approved accounts.
- Share the minimum needed, and swap names for placeholders.
- Never paste passwords, keys, payment, or health data.
- Know whether your plan trains on your inputs.
- Keep customer contracts and unreleased plans out unless approved.
- Report mistakes quickly. Early reports are cheap.
Earn customer trust
- Be relevant, not creepy. Reference professional context only.
- Keep your own voice and real opinions in the message.
- Disclose AI use where law or contract requires.
- Never claim to be human if someone sincerely asks.
- Honor unsubscribe and do-not-contact requests.
- Never invent references, quotes, or urgency.
Build habits
- Start with one task and time-box it.
- Keep a personal library of prompts that worked.
- Use AI for the first draft, not the last word.
- Keep a short log of wins and misses.
- Review automations on a schedule, with a sample audit.
- Share what you learn with your team.
For managers
- Publish simple rules using the three traffic lights.
- Create a shared, approved prompt library.
- Measure outcomes such as response quality and time saved, not usage alone.
- Make it safe to admit mistakes.
- Keep a direct line to Legal and IT.
- Coach the checking habit as much as the prompting habit.
Before you hit send
Compare the AI tools
There are many AI tools, and each comes from a company with different strengths. Pick the ones you want side by side. These are directional profiles, not benchmarks. Model names, plans, and data terms change often, so confirm every row with the vendor and your IT team.
Which one fits where my team already works?
Use case library
Each card is one task, with a light: green is safe, amber needs guardrails, red is not safe. "Safe" assumes you use a tool your company has approved and you review the output. Nothing here is permission to use a tool, so confirm with Legal and IT.
Three questions before any prompt
Answer these in order. One bad answer is enough to stop or escalate.
1. What data goes in?
- Public: websites, filings, press releases. Lowest risk.
- Internal: playbooks, non-public pricing, forecasts. Approved enterprise tool only.
- Customer or regulated: contracts, personal data, health, payment. Needs written approval from Legal and IT.
2. Who checks the output?
- A named person reads everything before it reaches a customer.
- Numbers, quotes, names, and claims get verified against a source.
- Nothing sends, saves to the CRM, or commits to terms on its own.
3. What if it is wrong?
- Low: a clumsy draft only you see.
- Medium: a wrong fact in a customer email.
- High: a pricing, legal, security, or compliance statement. Needs expert sign-off.
Safe Green
Public or your own non-confidential content, approved tool, you review. Go.
Guardrails Amber
Touches internal or customer data, or affects customers. Allowed only with the control listed on the use case and sign-off from Legal and IT.
Not safe Red
Breaks confidentiality, privacy law, consent, or honesty. Do not do it, even in an approved tool, unless Legal explicitly says otherwise.
Laws and rules that touch sales AI
A plain-language overview, not legal advice. Rules differ by country and state and change often. Ask Legal what applies to you.
Laws such as CAN-SPAM in the US, CASL in Canada, and ePrivacy rules in the EU and UK cover commercial email. Common themes: honest sender and subject lines, a working opt-out, and in some places prior consent. AI-written volume does not change the rules.
GDPR, UK GDPR, CCPA/CPRA, and many state laws limit how personal data is collected and used. Putting personal data into an AI tool is processing it, which can require a lawful basis, vendor terms, and transparency.
Some places require everyone's consent and others only one party's. AI note-takers and transcription count as recording. Ask Legal about every region you sell in.
In the US, the TCPA restricts autodialed and prerecorded calls and texts, and regulators have said AI-generated voices count as artificial voices. Consent rules are strict.
Some laws and contracts require telling people when they are talking to an AI or when content is AI-generated, such as the EU AI Act's transparency duties and several US state chatbot laws.
NDAs and customer agreements often limit sharing information with third parties, which can include AI vendors. Check before pasting anything covered.
Health (HIPAA), financial services (GLBA and securities rules), payments (PCI), and public-sector buyers add stricter requirements. Ask before selling or using AI there.
AI output can resemble existing work. Check images, text, and music before using them publicly, and follow your brand and approval rules.
If you slip
Pasted something you should not have? It happens. Fast, honest reporting is what keeps a small mistake small.
Agents and connected tools: extra care
AI agents can read your email, browse the web, and take actions. That is powerful, and it is a new kind of risk.
- Hidden instructions are real. A web page or email can contain text that tries to steer the AI. This is called prompt injection.
- Give the least access that works. Start read-only. Do not grant send, delete, or payment rights early.
- Keep a person approving outside actions. Sending an email or changing a record should wait for a human.
- Never put secrets in prompts or connected files. Passwords and keys belong in a vault.
- Know how to turn it off. Have an off switch and a log of what the agent did.
- Ask IT first. Connecting tools to your accounts needs security review.
Rules that apply to every vendor
- The plan matters more than the brand. Free and personal accounts often have different data terms than business plans.
- Use work accounts only. Never move work data through a personal account.
- Check call recording and transcription consent rules in every region you sell in.
- Disclose AI where law or contract requires it, and never deny being an AI when someone sincerely asks.
- Ask for the vendor's data retention, training, and subprocessor terms in writing.
Check your instincts
Ten real sales situations, drawn at random from a larger set, so a retake is different. Tap an answer to see why. Your best score is saved on this device.
Prompt library
Copy a prompt, replace the [brackets], and read the result critically. Use only public or your own content, in a tool your company approved. Each prompt lists what to check before you use the answer.
Playbook
Worked examples, a spot-the-error exercise, what to say to customers, and answers to common rep worries. All names here are fictional.
Example: from generic to useful
"Dear Sir/Madam, I hope this email finds you well. Our industry-leading, best-in-class solution can revolutionize your business. We have helped hundreds of companies achieve amazing results. Let me know when you are free for a call."
"Hi Priya, I saw Northwind is hiring an operations lead. When that role opens, keeping scheduling consistent across plants is often the first thing to slip. Is that on your list this quarter? I can share a one-page example of how another fabricator handled it. Sam"
- Specific reason to write. A public trigger replaces flattery.
- No unproven claims. "Hundreds of companies" and "revolutionize" are gone.
- One clear question. The reader knows what to do next.
- You verified it. Only claim "another fabricator" if it is true and you can share it.
Example: upgrade the prompt
Summarize this call.
Summarize the call notes below for my manager in under 120 words. List the customer's top three priorities, any objections, and next steps with owners and dates. Mark anything unclear as "unclear". [notes with names replaced by roles]
Exercise: spot what to verify
You pasted this public blurb about a fictional company: "Northwind Fabrication makes custom metal parts for construction and sells to regional contractors. A recent press release mentions hiring an operations lead." The AI returned the brief below. Tap every sentence you would verify before using it.
What to say to customers
Starting points only. Use the wording Legal approves for your company.
"Fair question. Here is exactly how we handle customer data: [approved statement]. If you need more detail, I will connect you with our security team."
Never: guess or say "no" without confirming.
"I used AI to help draft it, and I reviewed and edited every line. Happy to tell you more about how I work."
Never: deny it when someone sincerely asks.
"I would like to use an AI note-taker so I can focus on you. It records and transcribes this call. Is that okay with everyone?"
Never: start without consent where it is required.
"Here is what it does well, here is where a person should review, and here is how we measure it: [approved facts]."
Never: promise accuracy you cannot prove.
Answer the real question inside it. Reply short and human, and ask one clarifying question.
Tip: AI polish does not mean low intent.
"That is a legal question, so I would rather connect you with the right expert than guess."
Never: pass AI answers off as legal, tax, or compliance advice.
Questions reps ask
Will AI replace me?
AI is good at drafts, summaries, and patterns. Buyers still choose people they trust, and trust, judgment, and relationships stay human. Reps who use AI well will likely outperform those who do not, so learning it is the safest career move.
Will I be judged on how much I use it?
That is a question for your manager. Healthy teams measure outcomes, such as quality and time saved, rather than raw usage.
What if the AI is wrong and I did not notice?
It happens to everyone. Follow the checking habits, report fast if something went out, and fix it with the customer. See "If you slip" on the Safety tab.
Can I use my own free account because it is faster?
Not for work data. Personal accounts often have weaker data terms and no company visibility. Ask IT for an approved option.
Do I have to tell customers I use AI?
Sometimes the law or the contract requires it, and you should never deny it when sincerely asked. Ask Legal what applies to your company.
My manager asked me to do something marked red. What now?
Pause and ask Legal or IT before doing it. Raise it calmly and put the question in writing. That protects you and the company.
Is using AI cheating?
No. It is a tool, like a CRM or a calculator. What matters is honesty, accuracy, and using approved tools.
How do I get better at it?
Pick one task a week, keep the prompts that work, compare results, and share what you learn with teammates.
Light Sorter
A retro sorting game. Real sales AI use cases fall from the top. Send each one to GO, CAUTION, or STOP before it lands. Every answer teaches you why.
How to play
- GO: safe, in an approved tool, with your review. CAUTION: allowed only with the control listed. STOP: do not do it.
- Assume every task uses a tool your company has approved.
- Keep a streak for a bigger multiplier. A level up comes every five correct answers and the cards fall faster.
- Ask Legal and IT crosses out one wrong answer. You get three per run.
- Chill mode removes the timer and the lives, so you can learn at your own pace.
Rollout checklist
Work top to bottom. Progress saves in this browser only.