By the end of this lesson you will be able to
- Cut prospect research prep from an hour to ten minutes
- Write follow-ups that reference something real instead of "just checking in"
- Know which part of selling to keep away from AI entirely
Selling is relationship work wrapped in an enormous amount of typing. The typing is the part to hand over — and doing so buys back the time that makes the relationship work possible.
Where it fits
| Task | Fit | Why |
|---|---|---|
| Turning call notes into a summary and next steps | Excellent | You supplied the material |
| Drafting a follow-up referencing the conversation | Excellent | Grounded in what you paste in |
| First draft of a proposal from your template | Very good | Structure is the hard part; it is good at structure |
| Prepping questions from a prospect's own website | Good | Only if you paste the page in |
| Objection handling practice | Good | Useful sparring partner, no stakes |
| "Tell me about this company" | Poor | Recall from training. It will confidently invent |
| Reading the room on a call | No | Yours. Entirely. |
Prospect prep in ten minutes
The pattern that works: gather real material yourself, then have AI turn it into a briefing. Gathering is fast. Reading and synthesising is what took the hour.
I have a discovery call tomorrow with [Name], [Title] at [Company].
Below is their about page, two recent news items, and [Name]'s
LinkedIn summary.
[PASTE MATERIAL]
Give me:
1. Three sentences on what they do, in plain language
2. Two things that have changed for them recently
3. Five questions worth asking, based on the above — not generic
discovery questions
4. One thing I should NOT assume about them
Keep it to one page.Follow-ups that are not "just checking in"
Most follow-ups fail because they contain no information. A follow-up that references something specific from the conversation gets replies; one that says "circling back" does not.
Here are my notes from a call with [Name] at [Company]
three days ago:
[PASTE ROUGH NOTES]
Draft a follow-up email from me.
Format: 4 sentences, no bullet points, no subject line fluff.
Constraints:
- Reference one specific thing they said, not a generic summary
- Include the one next step we agreed
- Do not restate the price
- Do not use the words "circling back", "touching base", or
"just checking in"Proposals: draft, do not delegate
AI is very good at getting a proposal to eighty percent from your template plus your notes. It is not good at the twenty percent that wins — the part where you show you understood their actual problem.
Give it your best previous proposal
Structure, tone, and section order all come from the example. This is far more effective than describing what you want, and it keeps your voice rather than producing generic vendor-speak.
Give it the discovery notes
The specifics are what make it theirs. Without them you get a template with the name changed, which is worse than no proposal because it signals you were not listening.
Rewrite the opening yourself
The first paragraph is where you demonstrate understanding. Write that one. Let AI handle scope, deliverables, timeline, and terms — the sections where consistency matters more than voice.
Check every number
Prices, dates, quantities, names. A proposal with a wrong figure is worse than a late one.
The CRM problem
Nobody updates the CRM because writing up a call after the call is the least appealing ten minutes of the day. This is a near-perfect AI task: rough notes in, structured summary out, paste into the record.
Turn these rough call notes into a CRM entry.
[PASTE NOTES]
Format:
- Summary (2 sentences)
- Their stated problem, in their words where possible
- Budget signals (or "none given")
- Decision process and who else is involved
- Next step with a date
- My honest read on likelihood, and why
Flag anything I clearly did not ask about that I should have.What to keep
- The call itself. Reading hesitation, hearing what someone did not say, deciding to stop talking.
- The judgement call on whether a deal is real. AI will happily rationalise a prospect who is never going to buy.
- Anything the customer will read as personal. A handwritten-feeling note lands because it took effort; the version that did not is usually detectable.
- The decision to walk away. That is a business call, and it needs someone accountable for it.
Should I tell prospects AI drafted the email?
For a follow-up you reviewed and edited, no more than you would disclose using a template. If AI produced analysis or a recommendation the client will act on, that is different — say so. The line is roughly whether they are relying on it.
Can it write cold outreach?
It can, and the output is usually detectably generic because it has nothing specific to work from. Cold outreach lives or dies on one true, specific observation about the recipient — which you have to find. AI can then shape it.
What about AI features in our CRM?
Often genuinely useful for summarising and drafting, because the data is already there. Check the data terms though — an AI feature inside a tool you already pay for may operate under different terms than the base product.
Key takeaways
- Gather real material yourself, then let AI synthesise it. Never ask it to recall a company.
- Ask what you should not assume — it surfaces where the material is thin.
- Follow-ups need one specific reference to the conversation. Ban "circling back" in the prompt.
- Proposals: AI does scope and terms, you write the opening paragraph.
- Rough notes to CRM entry is the highest-return task in the list, and the one nobody does.