Six playbooks, in the order they usually matter

Your first week with Claude, without changing anything else

1. Day 1. Take the last email you wrote by hand and ask Claude to rewrite it so the first sentence is about the recipient. Compare. Keep whichever is better.
2. Day 2. Before your first meeting, paste your notes from the previous call and ask for the four-point prep brief. Notice what you had forgotten.
3. Day 3. Paste your current sequence and ask which email you should delete. Delete it.
4. Day 4. Take a call recording transcript and ask for your own talk-to-listen ratio. Do not share it with anyone.
5. Day 5. Pick the one prompt from the week that saved real time. Save it somewhere you will find it Monday.

The call prep routine that survives a full calendar

1. Keep one saved prompt, not five. Prep is the highest-frequency job, so it is the one worth making frictionless.
2. Feed it the CRM record and the last email thread. If you are pasting these by hand every time, that is the signal to connect the data properly rather than to write a better prompt.
3. Ask for one thing you have been avoiding asking the account. This single line finds more risk than the rest of the brief.
4. After the call, spend 60 seconds pasting your rough notes back and asking for the follow-up draft. The draft is worse if you wait until evening.

Account research that does not repeat the homepage

1. Ask for what is not on the website: hiring patterns, tooling overlaps, what changed in the last quarter, who is new in the buying centre.
2. Give the model your ICP in writing before asking it to judge fit. Without that, fit means whatever the model assumes it means.
3. Demand the source for each claim. Anything unsourced is a guess wearing a suit.
4. Finish with one line: 'which single fact here justifies reaching out this week rather than next quarter'. If there is no answer, the account is not ready.

Rolling it out to a team without a mandate

1. Week 1. Two volunteers, one workflow, no announcement. Pick the two reps who already experiment.
2. Week 2. Those two demo their actual output in the team meeting. Real emails, real prep, including where it was wrong.
3. Week 3. Publish the three prompts that survived contact with the pipeline. Do not publish a library of forty.
4. Week 4. Connect the data source everyone is pasting by hand. This is the step that turns a novelty into infrastructure.

Guardrails worth setting before a rep sends anything

1. Drafts, never sends. Every email workflow ends in the drafts folder for the first month at minimum.
2. No compliance, security, legal, or pricing claim leaves a chat window without a human who owns that area confirming it.
3. Never paste a customer list, contract, or anything under NDA into a personal account. Use the company workspace, which has the retention terms your legal team agreed to.
4. Write down what good output looks like for your team, with one example. Otherwise every rep calibrates differently and quality debates go in circles.

The point where pasting stops scaling

1. You notice you paste the same CRM record every morning.
2. Your prep quality depends on whether you remembered to include the email thread.
3. Two reps get different answers for the same account because they pasted different context.
4. You want yesterday's calls summarised into the CRM and there is no version of pasting that does it.

Auditing where you actually are

Teams tend to overestimate adoption because two enthusiastic reps make the whole team look active. A checklist beats a survey here. Salesgear's AI sales checklist is 30 items covering data, tools, process, and enablement, and it is deliberately hard to score well on.

The item most teams fail is the last playbook above. Everything works in the demo because someone pasted perfect context by hand, and nothing works in month two because nobody has time to do that daily. Fixing that is a different kind of project, and it is the one that determines whether any of this outlives the quarter.