Option A: ChatGPT: groups your correspondence into a checklist
Best when: You want candidates from correspondence, not memory
What it does: ChatGPT groups and builds; you check each candidate
Strong at: Groups, de-duplicates and tables in one chat
Limit: Access, limits and accuracy not covered by cited page
Free plan: Yes, limited (free tier limits images, voice, deep research, Codex)
Option B: Claude: guided memory interview by area of life
Best when: Correspondence is incomplete; use a guided memory interview
What it does: Claude asks questions; you answer from records
Strong at: One neutral question at a time keeps categories open
Limit: Accuracy not validated by the cited page
Free plan: Yes, limited (usage capped in 5-hour and weekly windows)
Option C: Gemini: work backwards from missed-mail risks
Best when: You want to start from missed-mail consequences
What it does: Gemini brainstorms risks; you match them to records
Strong at: Risk-first worksheet from your own circumstances
Limit: Accuracy not validated by the cited page
Free plan: Yes, limited