Should same-name contacts merge? Check field-level evidence

Supply synthetic contacts and merge rules to draft candidate matches, field conflicts and a recovery mapping without assuming same names identify one person.

Execution environment and usage

AI / software

Claude

Where to operate

The Claude web-chat input; this version requests a text draft.

Getting started

  1. Open an eligible account at claude.ai and start a chat without connecting external accounts.
  2. Prepare synthetic contact records, source identifiers and merge rules.
  3. Request the draft/check and compare it with the original material. Suggestions are not evidence of execution.

Setup and plan requirements

Task inputs: Prepare synthetic contact records, source identifiers and merge rules. Use an eligible tool that can read the supplied format; verify parsing, context limits and access before relying on it.

Prompt portability limits

The written request can be adapted to a chatbot that accepts the same input. It does not transfer the original agent, plugins or computer permissions.

Usage limitations

This supplied-material exercise does not reproduce the original automation. Verify output, calculations and sources; regional/account availability is not guaranteed.

Environment sources

適用情境 / When to use

Supply synthetic contacts and merge rules to draft candidate matches, field conflicts and a recovery mapping without assuming same names identify one person. Do not import real private address books. Do not rank job candidates or infer sensitive traits. Original case: The author demonstrates a workflow without measuring false-merge rates.

步驟 / Steps

  1. Prepare synthetic contact records, source identifiers and merge rules.
  2. Separate identity evidence, field conflicts and inferred information.
  3. Retain field provenance and a reversal mapping; keep unresolved identities separate.

提示詞 / Prompt

From [synthetic contact records/source identifiers/merge rules/field definitions], propose record merges. Preserve provenance per field and separate verified facts, conflicts and inferences; do not merge same-name records without sufficient evidence. Include a mapping to recover originals. Do not modify databases, score applicants or process private address books.

留意事項 / Cautions

  • Do not import real private address books.
  • Do not rank job candidates or infer sensitive traits.

來源限制 / Evidence limits

Similar names do not establish identity; inferred skills are not verified experience.

流程說明 / Workflow note

These steps and prompts are original, untested exercises by this site, not the author’s verbatim workflow. Reported results do not guarantee exercise outcomes.

原始工具組合 / Original tool stack

Claude; Supabase MCP; Harmonic; Granola

Source publication date: 2026-03-12

Source checked: 2026-10-20

Uncovered Bookmarks: Building a Personal CRM