Choose one of three approaches. The steps for that option appear below.
Choose an option above. On a phone, swipe the tabs left or right to see all options →
The brief, evidence and approved summary sit in one Claude project. Claude plans the arc, then drafts; you approve both.
Collect inputs
Plan three touches
Check the plan
Draft the touches
Check and approve
Use: You
Have ready: the account brief and personalisation evidence, and the approved offering summary.
Check the brief and summary are the current approved versions.
Use: Claude
Put the brief, evidence and approved summary in a Claude project and ask for a three-touch plan with purpose, evidence reference and spacing per touch.
What to paste into the tool
Task: Plan table: touch, purpose, spacing, evidence reference. Plan three touches: opener, value follow-up, clean close. For each give purpose, the exact brief or offering line it rests on, and a spacing suggestion. Use only capability claims in the approved summary.
Use: You
Check the arc matches the research and every claim is in the approved summary; ask for changes in the same chat.
Check each touch adds something new; no resending the same pitch.
Use: Claude
Ask Claude to draft all three touches from the approved plan, each with an evidence note and its purpose and spacing.
What to paste into the tool
Task: Three drafts with purpose, spacing and evidence notes. Draft the three touches from the approved plan only, keeping the codes in place; do not invent names. Under each, list the claims and their source lines. Do not add claims. Include each touch's purpose and spacing suggestion from its plan.
Use: You
Check each claim against the summary and brief, ask Claude to change only what you list, then approve each touch.
Nothing is sent, scheduled or uploaded to a sending tool by this workflow.
You write the three-touch plan. ChatGPT drafts each touch from its plan row; you approve.
Write the plan
Draft each touch
Read and approve
Use: You
Have ready: the account brief, personalisation evidence and approved offering summary. Write purpose, evidence reference and spacing per touch.
Check each touch adds something new; every claim is in the approved summary.
Use: ChatGPT
For each touch, in a fresh chat, paste its plan row and the matching lines and ask for a short draft with purpose and spacing.
What to paste into the tool
Task: Draft touches with purpose, spacing and evidence notes. Write touch [number] from this plan row and these lines only. List which line backs each claim. Do not add facts. Include each touch's purpose and spacing suggestion from its plan.
Use: You
Read the three drafts together, remove repeated pitch, then approve each touch.
Nothing is sent, scheduled or uploaded to a sending tool by this workflow.
Gemini Notebook proposes angles with supporting passages, you write the three touches, and ChatGPT audits claims and repetition.
Collect inputs
Propose angles
Write the drafts
Audit the drafts
Check and approve
Use: You
Have ready: the account brief and personalisation evidence, and the approved offering summary.
Check the brief and summary are the current approved versions.
Use: Gemini Notebook
Add the brief, evidence and summary as sources and ask for angles per touch with the supporting passage for each.
What to paste into the tool
Task: Angles with supporting passages. Keep the codes unchanged. Give the passage supporting each. Say when no passage supports an angle.
Use: You
Check each passage supports its angle. Pick one angle per touch, set purpose and spacing, and write the three drafts with evidence notes.
Check each touch adds something new; only approved capability claims.
Use: ChatGPT
Paste the drafts and approved summary and ask for any claim not in the summary and any touch that repeats another.
What to paste into the tool
Task: List of unsupported claims and repeats. List every claim in these drafts that is not in the approved summary, and any touch that repeats another. Keep the codes unchanged. Do not rewrite.
Use: You
Check every flag, read the unflagged claims yourself, then approve each touch. Nothing is sent, scheduled or uploaded to a sending tool by this workflow.
Check each touch is approved with its purpose, spacing suggestion and evidence notes.
Best when: You want plan and drafts in one workspace
What it does: Claude plans and drafts; you approve the plan, then the drafts
Strong at: Project uploads are seen across all chats in the project
Limit: Free users can create up to five projects
Free plan: Yes, limited (5-hour and weekly usage caps)
Best when: You want each draft tied to its own plan row
What it does: ChatGPT drafts each touch; you wrote the plan and approve
Strong at: Drafts from supplied lines and instructions
Limit: Three separate chats mean repeated pasting
Free plan: Yes, limited (file uploads, images, voice, data analysis)
Best when: You want to write the touches and have claims audited
What it does: Gemini Notebook cites angles, ChatGPT audits; you write and approve
Strong at: Answers from the sources you upload
Limit: Passage-per-angle not confirmed; check each passage yourself
Free plan: Gemini Notebook: Yes, limited (500,000-word sources); ChatGPT: Yes, limited (upload caps)
Before you paste into any AI tool, remove any confidential information. Put it back at the end.
Deciding between ChatGPT and Claude for this? See the comparison.
Every option gives each touch a distinct purpose with evidence, uses only approved capability claims, and ends with the seller approving every touch.
Check the result against your original inputs. Approve it before sharing or acting.