Practitioner
Your AI Plan: Occasional Experimenter
You already use AI. This month is about making it work the same way every time.
You scored between 10 and 17, which makes you an Occasional Experimenter. Your AI results swing because most of your work starts from a blank chat. You don't need more tools. You need three habits that stop each chat from starting at zero.
In 30 seconds
- Your profile: "You use AI for some tasks, but results are hit or miss and most work still starts from a blank chat. A saved context, structured prompts and a fact checking habit will make your results consistent."
- Save your context once, so you stop retyping it and stop forgetting it.
- Rebuild the prompt you use most, so the fixes you make every time become part of the prompt.
- Check by stakes, not by surprise. The wrong fact that hurts is the one that sounds right.
- Expect about 30 minutes a week for four weeks, then retake the assessment.
What Occasional Experimenter means (and what it doesn't)
Scroll sideways to see the whole table.
| It means | It doesn't mean |
|---|---|
| AI is in your week, mostly for tidying emails and quick questions | You've plateaued. The next step up is setup, not talent |
| You paste background "when you remember," so some chats know your product and some don't | The tool is unreliable. Most bad output is a context problem, not a tool problem |
| You keep asking for changes until a draft is close | Your prompts are bad. You're already giving feedback; it just isn't saved anywhere |
| You double check facts that seem surprising | You're careless. You have a checking instinct; it's aimed at the wrong facts |
Strengths people at this level usually have
- Real reps with AI. You already know which tasks it helps with and which it doesn't.
- An iteration habit. You don't accept the first draft, and specific feedback is half of good prompting.
- A saved prompt or two. If you keep prompts in a doc, you've already taken the first step toward a prompt template.
Your three biggest opportunities
Your results screen picked your three lowest areas. When scores tie, the order is Fact checking, Data safety, Workspace setup, then Prompt structure, so your three may differ slightly from the ones below. If Data safety is one of them, do it first. Below are the three your profile description names: the tip from your results screen, then what it means at your level.
Save your context (Workspace setup)
"Write a one page seller context file: what you sell, your ideal customer, key personas, your voice and your rules. Save it in a Project so every chat starts with it." At your level: you've already typed most of this across old chats. Gather those pieces instead of starting blank. How to Build a Seller Context File has the template.
Structure your go to prompt (Prompt structure)
"Rebuild your most used prompt in five parts: goal, context, inputs, rules and format. Test it on two real accounts and save the version that wins." At your level: list the edits you make every time ("shorter," "less formal," "bullets"). Each one belongs in the rules or format part. Prompting 101 explains all five.
Check by stakes (Fact checking)
"Open every source before a fact reaches a buyer. A link in the output is not proof until you click it, and pricing, security or ROI claims also need approval from the right owner." At your level: checking only what seems surprising misses the dangerous kind of hallucination: one that sounds normal, like a wrong job title.
Scroll sideways to see the whole table.
| Stakes | Examples | Your check |
|---|---|---|
| Low | Brainstorm, roleplay, a rough call plan | Skim for sense |
| Medium | Prospect email, LinkedIn message, call opener | Open the source for every fact about the prospect |
| High | Pricing, security, contract terms, ROI numbers | A source for every claim, plus approval from the right owner |
How AI Assistants Actually Work explains why AI writes guesses and facts in the same confident tone.
Your 30 day plan
Your results email sends one focus area each Monday for three weeks, then a final week to put it together. Here's how to use them at your level. If your focus areas came in a different order, swap the rows.
Scroll sideways to see the whole table.
| Week | Your email covers | What to do at your level | Time |
|---|---|---|---|
| 1 | Your first focus area | Turn the background from old chats into a one page seller context file saved in your workspace. Run the guide's tests | About 30 minutes |
| 2 | Your second focus area | Pick the prompt you type most. Rebuild it with all five parts. Run old and new on two real accounts, keep the winner and save it with {placeholders} |
About 30 minutes |
| 3 | Your third focus area | Run the stakes check before every prospect send. Keep a short log: which facts held up, which didn't, and where the wrong ones came from | 5 minutes per send, plus 10 on Friday |
| 4 | Put it together and measure | Run your saved prompt on three new accounts. Score each output 1 to 5 for accuracy, specificity and "usable with no edits." Time one run, then retake the assessment | About 30 minutes |
Prompts to start with
These work on material you already have. Run them in your company approved tool.
Prompt
Gather what you've already told AI
Below are pieces of background I've pasted into AI chats over the last few weeks: product notes, bits about my buyers, emails I liked. Organize them into a one page seller context file with six headings: About me, What we sell, Ideal customer profile, Personas, My voice and Rules. Keep my wording. Don't invent anything. At the end, list the 5 most important gaps you'd need me to fill.
<my_pieces>
{paste the pieces}
</my_pieces>Prompt
Rebuild your go to prompt
Here's a prompt I use a few times a week, and a recent output from it that I had to fix. Rewrite the prompt with five labeled parts: goal, context, inputs, rules and format. Turn every fix I made into a rule or a format instruction so I don't have to make it again. Before you rewrite, ask me up to 5 questions about anything you need.
<my_prompt>
{paste the prompt}
</my_prompt>
<output_i_fixed>
{paste the output and what you changed}
</output_i_fixed>Prompt
Stakes check before you send
Here's a draft I plan to send to {name}, {title} at {company}. List every factual claim in it. Label each one low, medium or high stakes: medium for any fact about the prospect, high for pricing, security, contract or ROI claims. For each medium or high claim, tell me where I could check it. Then tell me which claim you are least sure about, and why.
<draft>
{paste the draft}
</draft>Then open each source yourself.
Prompt
Diagnose a miss
Here's a prompt I ran and the output I got. The output missed. Tell me which of these caused it: (a) you lacked context about my product, buyer or voice, (b) my instructions didn't say what I wanted or how to format it, or (c) you filled a gap with a guess. Point to the exact lines that show it. Then tell me the one thing to add to my context file or my prompt so this miss doesn't come back.Quick check
Which fix does this miss need?
- The email could be from any company to any buyer. Reveal: Context. Add your product, buyer and voice to your context file.
- The content is right, but it's four paragraphs when you wanted three bullets. Reveal: Structure. Add a format line to your prompt.
- The brief says the prospect's VP of Sales is someone who left last year. Reveal: Fact checking. It sounded normal, so only a stakes check catches it.
Try it now
Your first 15 minutes
- Open your last few AI chats. Copy the background you typed: product details, buyer notes, emails you liked.
- Run the "Gather what you've already told AI" prompt on those pieces, in your company approved tool.
- Save it in a Project or similar saved workspace (names differ by tool; check its current documentation).
- Test it: "Using my context file, describe our product to a CFO in 3 sentences. Then tell me what's missing from my context that would make your answers better." Fill one gap now.
Your consistency checklist
0 of 7 done
Ticks are saved in this browser only.
How to use your personal focus areas
Your results showed three focus areas, weakest first. Work on them in that order, one per week, and leave the other seven for now. Each focus area will get its own page; until then, the free guide next to it covers the same ground.
- Fact checking · guide: How AI Assistants Actually Work · How to Use AI for Account Research
- Data safety · guide: AI Policy for Sales Teams
- Workspace setup · guide: How to Build a Seller Context File
- Prompt structure · guide: Prompting 101 · AI Prompts for Sales
- Outreach writing · guide: Cold Emails with AI
- Account research · guide: How to Use AI for Account Research
- Everyday use · guide: AI for Sales: A Practical Guide · goal: Get your mornings back
- Call prep and practice · guide: AI Sales Roleplay · goal: Stop freezing on objections
- Post call admin · guide: AI for Sales Call Notes
- Automation · guide: AI Sales Automation
Key idea
Hit or miss results are a setup problem, not a talent problem. Save what AI needs to know, and it stops guessing.
Retake the assessment
Retake it at day 30, answering for your normal week. Capable Practitioner starts at a score of 18. Each answer you move up one step adds a point.
Free assessment
What's next
You don't need to buy anything to make progress this month. If you want all ten areas with live practice, the certification fits this level.
Scroll sideways to see the whole table.
| Step | What it is | Cost |
|---|---|---|
| Now | How to Build a Seller Context File and Prompting 101 | Free |
| This month | This 30 day plan, plus AI Prompts for Sales | Free |
| When you want all ten areas | The AI for Sales Certification: 10 self paced sessions (about five weeks at 2.5 hours a week) with an optional weekly live class on your own accounts, ending with the Certified AI Sales Practitioner credential. You have the habit; the course turns it into a system | Paid |
| After that | Your AI Plan: Capable Practitioner, once your next score lands at 18 or above | Free |
Know your role? AI for Prospecting Reps and AI for Closing Reps show where these habits fit in a real selling week. Find your path gives you the full reading order.
Lead a team? If you took this as a manager, see team options or estimate the time your team could get back.
Frequently asked questions
Why are my AI results so inconsistent?
Because each chat starts with different information. When you forget to paste background, AI fills the gaps with generic guesses. A saved context file and a structured prompt make every chat start from the same place.
Is a score of 10 to 17 good?
It's a solid middle. AI is already part of your week. What's missing is setup and habits, which are quicker to fix than starting from scratch.
I already save prompts in a doc. Isn't that enough?
It's a good start, but a prompt without your context still gets generic answers. Pair it with a saved seller context file.
Should I take the certification now or wait?
Take it when you can protect about 2.5 hours a week for five weeks. Until then, finish this month first: the free plan on this page works on its own.
Why do my chats get worse the longer they run?
Because long threads fill the context window, and early instructions start to lose weight. If a chat is still off track after about four rounds of changes, start a fresh one with a better prompt and your saved context.
AI for Sales Certification
Start week 1 today
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
