Sales
Prep calls in ten minutes, follow up within the hour, keep the CRM current through connector writes you approve, and review pipeline every Monday, without inventing a customer claim.
For Account executives, SDRs, sales managers and revenue operations
Prep every call, follow up the same hour, and keep the CRM honest
- Call prep in about 10 minutes: account brief, deal history, agenda and discovery questions
- A follow-up email drafted within an hour of every call, with CRM changes proposed and evidenced
- CRM fields updated through connector writes you approve, not a Friday catch-up
- A Monday pipeline review that flags stale deals, past close dates and missing next steps
- Every figure in outbound email traced to an approved source
On this page
- Where AI helps in this job
- Start in 15 minutes
- The tools people in this role use
- Members only: Workflow: account research and call prep
- Members only: Workflow: follow-up and CRM update after every call
- Members only: Workflow: a proposal built only from approved claims
- Members only: Workflow: objection practice
- Members only: A skill for sales
- Members only: Guardrails and hooks
- Members only: Routines that run without you
- Members only: Prompt ladder for sales
- Members only: Measure it
- Members only: Risks and what not to automate
A sales week has the same shape every week: research accounts, prepare calls, run discovery, follow up, update the CRM and review the pipeline. AI is fast at the reading and the first drafts. It does not know what the buyer meant, and when it has no proof it writes proof that sounds right. You own every word that reaches a customer.
Where AI helps in this job#
| Task | What AI does well | What you still own |
|---|---|---|
| Account research | Reads the company site, filings, news and job posts and writes a one-page brief with links | Which facts matter for this deal, and opening the links |
| Call prep | Pulls CRM history, recent email and past call notes into one brief with an agenda | The goal of the call and what you will ask for |
| Discovery questions | Drafts open questions tied to the account's likely priorities | Listening, and following the answer instead of the list |
| Follow-ups | Turns a transcript or notes into a recap email with owners and dates | Tone, commitments, and pressing send |
| CRM hygiene | Proposes stage, next step and close date updates, each with the quote that supports it | Approving each change, and the forecast category |
| Proposals | Assembles a first draft from discovery notes, approved pricing and approved case studies | Pricing, terms and every claim about results |
| Objection practice | Plays a skeptical buyer and scores your answers against a rubric | Judgment in the real conversation |
| Pipeline review | Flags stale deals, past close dates, single-threaded deals and missing next steps | The call on which deals are real |
AI is weak at reading the room, at knowing what was said off the record, and at telling a buying signal from politeness. It is often wrong about private companies: headcount, funding and tech stack on the web go stale fast. The bigger risk is specific to sales. Ask a model for a persuasive email and it adds a customer result or a percentage that sounds right. That is how invented claims reach buyers, and in the US the FTC's 2024 rule makes fake testimonials illegal, including AI-generated ones attributed to people who do not exist.
Start in 15 minutes#
These work in Claude, ChatGPT, Gemini or Microsoft Copilot (formerly Microsoft 365 Copilot). Turn on web search for the first one.
Prompt
One-page account brief
Research [COMPANY] ([WEBSITE]) before my first call with their [ROLE OF THE PERSON I AM MEETING]. Use web search and link every fact to its source. Write one page with these sections: 1. What they do and how they make money, in two lines. 2. Events from the last six months (funding, launches, leadership changes, layoffs, acquisitions), each with a date and a link. 3. Three things a [ROLE] at a company like this is likely measured on, and why you think so. 4. Three reasons our [WHAT WE SELL, IN ONE LINE] could matter to them. Label these as guesses. 5. What you could not find or confirm. Mark any source older than 12 months as possibly stale. Do not guess headcount, revenue or tech stack.
Good output links every line in sections 1 and 2. Open each link before the call and delete any line you cannot confirm.
Prompt
Discovery questions for a first call
I have a first call with the [ROLE] at [COMPANY], a [INDUSTRY] company with about [NUMBER] employees. We sell [WHAT WE SELL, IN ONE LINE]. Write eight open discovery questions that uncover how they handle [PROBLEM WE SOLVE] today, what it costs them, who decides, and their timing. No yes or no questions, and nothing they would expect me to know from their website.
Good questions start with how, what or walk me through, and only the buyer can answer them. Cut any question that pitches, keep four, and put them in your own words.
Prompt
Follow-up email from my call notes
Here are my notes from today's call with [NAME, ROLE] at [COMPANY]: [PASTE NOTES] Write a follow-up email under 150 words. Open with their top priority in their own words. List the next steps we agreed, each with an owner and a date. End with one question that moves the deal forward. Use only facts from my notes. Do not add product claims, percentages, customer names or results. If a next step has no date in my notes, write "date to agree" instead of inventing one.
Read each sentence against your notes and cut anything you cannot point to. The "date to agree" markers are deliberate. Agree the date with the buyer, then fill them in.
Warning
Buyers share confidential information with you: pricing, contracts, internal plans, org charts. Keep it in the workspace your company approved for customer data, not a personal chat account. Do not use scripts, browser extensions or agents to scrape LinkedIn or to automate actions there. Sections 8.2.2 and 8.2.13 of the LinkedIn User Agreement prohibit both.
The tools people in this role use#
- Your general assistant. Claude, ChatGPT, Gemini or Microsoft Copilot for research, briefs, drafts and role-play. Use the one your company approved.
- CRM connectors. The HubSpot connector for Claude reads contacts, companies, deals and activity history, creates or updates contacts and deals, and logs notes and tasks. It cannot delete records, writes at most 10 records per bulk change, and needs a paid Claude plan. HubSpot's remote MCP server, generally available since April 2026, gives other MCP clients similar access. Salesforce in Claude is in beta for organizations Salesforce approves, sees only what your Salesforce permissions allow, and asks you to approve each change by default.
- Sales agent in Microsoft Copilot. CRM context inside Outlook and Teams for Dynamics 365 Sales or Salesforce Sales Cloud (Professional edition and above). It needs a Microsoft Copilot licence, which Microsoft sold as Microsoft 365 Copilot before the rename.
- Gong MCP server. Lets Claude, ChatGPT or Microsoft Copilot ask about an account or deal and generate a brief from Gong calls and emails. It is read-only, leaves out private calls, and a Gong tech admin turns it on.
- Claude Cowork with Anthropic's sales plugin. 36 open-source skills, including call-prep, call-summary, pipeline-review, crm-hygiene-check and handle-objection, that work with whichever CRM, inbox and call recorder you connect, or with uploaded exports.
- Agentforce Sales Coach for Salesforce teams: deal-specific role-plays built from CRM data, with feedback on tone, clarity and objection handling.
Workflow: account research and call prep
Members only below this line
Get the full Sales playbook free
What you get on this page:
- 4 step-by-step workflows
- 13 prompts
- 1 skill
- 1 hook
- 2 routines
Also in the members part
- Workflow: follow-up and CRM update after every call
- Workflow: a proposal built only from approved claims
- Workflow: objection practice
- A skill for sales
- Guardrails and hooks
- Routines that run without you
- Prompt ladder for sales
- Measure it
- Risks and what not to automate
Sources
Checked against these on 13 Oct 2026.
- HubSpot: Set up and use the HubSpot connector for Claude (opens in a new tab)knowledge.hubspot.com
- HubSpot changelog: Remote HubSpot MCP server is now generally available (April 2026) (opens in a new tab)developers.hubspot.com
- Claude Help Center: Use Salesforce in Claude (opens in a new tab)support.claude.com
- Microsoft Learn: Microsoft Copilot requirements (rename note) (opens in a new tab)learn.microsoft.com
- Microsoft Learn: Sales agent FAQ (opens in a new tab)learn.microsoft.com
- Gong Help: About Gong MCP server (opens in a new tab)help.gong.io
- Anthropic knowledge-work-plugins: sales plugin (opens in a new tab)github.com
- Claude Help Center: Use connectors (tool permissions) (opens in a new tab)support.claude.com
- Claude Help Center: Schedule recurring tasks in Claude Cowork (opens in a new tab)support.claude.com
- Claude Help Center: Use skills in Claude (opens in a new tab)support.claude.com
- Claude Code docs: Hooks reference (opens in a new tab)code.claude.com
- Claude Code docs: Routines (opens in a new tab)code.claude.com
- FTC: Final rule banning fake reviews and testimonials (August 2024) (opens in a new tab)ftc.gov
- FTC: CAN-SPAM Act compliance guide for business (opens in a new tab)ftc.gov
- LinkedIn User Agreement (section 8.2) (opens in a new tab)linkedin.com
- Salesforce: AI sales coaching with Agentforce (opens in a new tab)salesforce.com
- Microsoft Support: Schedule your most used Copilot prompts (opens in a new tab)support.microsoft.com
- Gemini Apps Help: Schedule actions in Gemini Apps (opens in a new tab)support.google.com