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 →
Claude: links repeat contacts and drafts the report
You paste the ticket histories and the repeat-contact definition. Claude groups contacts and drafts the report. You reconcile every count.
- You
Prepare the ticket bundle
- Claude
Group contacts, draft report
- You
Reconcile and finish
Have ready: the ticket histories and definition
Use: You
Take credentials and payment details out of the ticket histories; bundle the histories with the supplied repeat-contact definition; keep the untouched originals aside.
Check
Check the bundle holds only supplied, approved material, with no passwords, authentication codes, session tokens or payment credentials.
Group contacts, then draft the report
Use: Claude
In one Claude chat, ask for the coding worksheet of repeat-contact sets, then the pattern report with counterexamples and unresolved questions.
Details
What to paste into the tool
Task: Draft coding worksheet of repeat-contact sets, unresolved asks per set, ticket references, counterexamples and unresolved gaps. From the attached histories, group contacts into repeat sets using exactly the supplied repeat-contact definition. For each set list the unresolved asks and the ticket IDs behind every count, and keep any contact that breaks the pattern as a counterexample. Do not infer motives or demographics; record anything the tickets do not show as a gap.
Task: Draft repeat-contact pattern report with evidence and counterexamples; Explicit unresolved questions and a proposed next human decision, not an executed action. Using only the reconciled worksheet pasted below, write the pattern report: each recurring friction pattern, the reconciled counts and ticket references behind it, the counterexamples, and the unresolved asks. Present proposed causes as possible causes, not confirmed. Close with unresolved questions and the proposed next human decision.
Reconcile every count and finish
Use: You
Reconcile every count and ticket reference against the retained originals, correct in the same chat, and close the report as preparation only.
Check
Check repeat counts reconcile against the originals and proposed causes read as possible causes, not confirmed. Do not resolve missing or contradictory facts by inference. Settle recipient scope and authority for later use separately; preparation only.
ChatGPT: drafts the coding sheet and the report
ChatGPT groups repeat contacts and rebuilds unresolved asks, then drafts the report. You reconcile every count against the originals.
- You
Prepare the ticket bundle
- ChatGPT
Group contacts, draft report
- You
Reconcile and finish
Have ready: the ticket histories and definition
Use: You
Take credentials and payment details out of the ticket histories; bundle the histories with the supplied repeat-contact definition; keep the untouched originals aside.
Check
Check the bundle holds only supplied, approved material, with no passwords, authentication codes, session tokens or payment credentials.
Group contacts, then draft the report
Use: ChatGPT
In one ChatGPT chat, ask for the draft coding sheet of repeat-contact sets, then the pattern report from the checked sheet only.
Details
What to paste into the tool
Task: Draft coding sheet of repeat-contact sets, unresolved asks per set, ticket references, counterexamples and unresolved gaps. From the attached histories, group contacts into repeat sets using exactly the supplied repeat-contact definition. For each set list the unresolved asks and the ticket IDs behind every count, and keep any contact that breaks the pattern as a counterexample. Do not infer motives or demographics; record anything the tickets do not show as a gap.
Task: Draft repeat-contact pattern report with evidence and counterexamples; Explicit unresolved questions and a proposed next human decision, not an executed action. Using only the checked coding sheet pasted below, write the report: each recurring friction pattern, the counts and ticket references behind it, the counterexamples, and the unresolved asks. Present proposed causes as possible causes, not confirmed, and do not infer customer motives or demographic traits. Close with unresolved questions and the proposed next human decision.
Reconcile every count and finish
Use: You
Reconcile every count and ticket reference against the retained originals, correct in the same chat, and close the report as preparation only.
Check
Check repeat counts reconcile against the originals and proposed causes read as possible causes, not confirmed. Do not resolve missing or contradictory facts by inference. Settle recipient scope and authority for later use separately; preparation only. Check the result against your original inputs. Approve it before sharing or acting.
Notion AI: groups a contact database and drafts
You load ticket histories into a Notion database. Notion AI groups repeat contacts and drafts the report. You reconcile counts.
- You
Load the contact database
- Notion AI
Group contacts, draft report
- You
Reconcile and finish
Load the histories into a Notion database
Use: You
Take credentials and payment details out of the histories; load one row per contact with its ticket reference, keeping the originals aside.
Check
Check the bundle holds only supplied, approved material, with no passwords, authentication codes, session tokens or payment credentials.
Group contacts, then draft the report
Use: Notion AI
Group rows into repeat-contact sets under the supplied definition noted on the page; then generate the pattern report as a doc from the grouped database.
Details
What to paste into the tool
Task: Grouped database of repeat-contact sets, unresolved asks per set and flagged counterexamples. Group the rows of this database into repeat-contact sets using exactly the supplied definition noted on the page. For each set, list the unresolved asks and flag any row that breaks the pattern as a counterexample. Do not infer motives or demographics; leave anything the rows do not show as a gap.
Task: Draft repeat-contact pattern report with evidence and counterexamples as a workspace doc; Explicit unresolved questions and a proposed next human decision, not an executed action. From the reconciled database, draft the report as a new doc: each recurring friction pattern, the reconciled counts and ticket references behind it, the counterexamples, and the unresolved asks. Present proposed causes as possible causes, not confirmed. Close with unresolved questions and the proposed next human decision.
Reconcile every count and finish
Use: You
Reconcile every set, count and counterexample against the retained originals, correct the rows, and close the report as preparation only.
Check
Check repeat counts reconcile against the originals and proposed causes read as possible causes, not confirmed. Do not resolve missing or contradictory facts by inference. Settle recipient scope and authority for later use separately; preparation only. Check the result against your original inputs. Approve it before sharing or acting.
Compare the three approaches
Option A: Claude: links repeat contacts and drafts the report
Best when: One analyst owns the review and wants one chat
What it does: Claude groups and drafts; you reconcile every count
Strong at: One chat carries the worksheet into the draft
Limit: Usage limits can pause a long session mid-work
Free plan: Yes
Option B: ChatGPT: drafts the coding sheet and the report
Best when: Count accuracy matters more than drafting speed
What it does: ChatGPT groups and drafts; you reconcile every count
Strong at: Synthesis over uploaded documents is a listed task
Limit: Free plan upload allowance is limited and unquantified
Free plan: Yes
Option C: Notion AI: groups a contact database and drafts
Best when: You want the coding to stay queryable after the report
What it does: Notion AI groups and drafts; you reconcile every count
Strong at: The setup persists and re-runs on a revised definition
Limit: Free plan gives only an AI trial
Free plan: Paid plan needed
Before you paste into any AI tool, remove any confidential information. Put it back at the end.
Final check
Never paste credentials or full payment details.
Suggestions and preparation only. Do not send, publish, change records or accounts, assign work, execute code, initiate a purchase, or commit a remedy or deadline. A human separately approves any downstream action.
Use only the supplied case evidence and approved rules; preserve source references, applicability dates and unresolved facts.
Cut down customer information in drafts and handoffs; do not request or reproduce passwords, authentication codes, session tokens or payment credentials.
If required source material or a qualified reviewer is missing, return a clearly labelled incomplete draft rather than inventing a result.
Do not infer customer motives or demographic traits; distinguish observed patterns from causal explanations and retain counterexamples.
Check the result against your original inputs. Approve it before sharing or acting.