{ "platformId": "chatgpt-work", "taskSlug": "task-2578", "directAnswer": "Use ChatGPT Work to convert authorized call notes or transcripts into a review-ready interaction record: preserve the customer's material words, reconstruct the event sequence, separate observed facts from summaries and interpretations, map each action to its actor and status, and surface the follow-up that remains open. Keep drafting separate from the customer record. ChatGPT must not write directly to the system of record, backdate an event, overwrite an earlier note, mark an action complete without evidence, or contact the customer. The attached usage estimates are projected role-task priors rather than measured ChatGPT use by customer service representatives, so they support a supervised trial, not a statement about adoption.", "topUseCases": [ { "interventionId": "task-2578:customer-interaction-to-reviewed-draft:approval-required", "taskId": "task-2578", "task": "Keep records of customer interactions or transactions, recording details of inquiries, complaints, or comments, as well as actions taken.", "workflowId": "customer-interaction-to-reviewed-draft", "authorityMode": "approval-required", "centralityScreen": { "value": 0.3662340406, "importance": { "value": 0.8825000000000001, "sourceField": "role-tasks-kv:task-2578.importance" }, "occupationalRelevance": { "value": 0.8220999999999999, "sourceField": "role-tasks-kv:task-2578.relevance / 100" }, "frequency": { "kind": "modal-response-share", "modalResponseSharePct": 50.48, "collectionDate": "2026-06-03", "sourceField": "role-tasks-kv:task-2578.modalResponseSharePct" } }, "agentFit": { "value": 0.9025, "sourceField": "rule:digital-feasibility-and-capability-fit-single-count.v2" }, "distribution": { "value": 0.9, "sourceField": "roles-kv:aiPlaybook.delegatedTasks[task-2578].observed-distribution" }, "evidenceConfidence": "weak", "withinGateScore": 0.29747359947734997, "gate": { "status": "approval-required", "reason": "A person must approve the verifiable handoff before action." }, "capabilityIds": [ "connected-apps-files", "deliverable-generation" ], "expectedArtifact": "Review-ready interaction record with attributed chronology, verbatim quotations, action-status ledger, and correction-safe provenance", "humanReviewPoint": "Verify identity, chronology, quotations, classification, and action status before entering the record; approve every later correction as a new attributable amendment.", "riskLevel": "medium" } ], "workflows": [ { "id": "customer-interaction-to-reviewed-draft", "goal": "Turn one customer interaction into an attributable, chronology-safe record draft without changing the system of record.", "inputs": [ "Authorized recording transcript or contemporaneous notes", "Interaction date, channel, participants, and verified customer or case identifiers", "Current record fields, reason codes, disposition taxonomy, and required disclosures", "Read-only prior case context and action history" ], "connectedSystems": [ "Approved transcript or note source", "Read-only customer case view", "Draft queue with no CRM write or send permission" ], "steps": [ "Verify the interaction identity, channel, participants, and source times; stop if the source could belong to another customer or case.", "Create an event chronology that preserves source order and distinguishes event time, statement time, note time, and proposed follow-up time.", "Populate required fields while keeping verbatim customer quotations, representative statements, system facts, summaries, and interpretations in separate labeled fields.", "Build an action ledger with actor, requested action, promised action, completed action, evidence of completion, owner, due condition, and current status; never infer completion from intention.", "Compare the draft to prior case context for contradictions, duplicate records, reopened issues, or promises that require escalation; do not overwrite the earlier record.", "Return the record draft, missing-field list, contradiction list, and action ledger; stop before CRM entry, disposition change, case closure, or customer contact." ], "expectedArtifact": "Interaction-record draft with source receipt, event chronology, labeled quotations and interpretations, action ledger, and exception list", "starterPrompt": "Turn this authorized transcript or contemporaneous note into a review-ready interaction record using the supplied schema and read-only case context. First verify customer or case identity, channel, participants, and source times. Preserve material customer quotations verbatim. Separate customer statements, representative statements, system facts, summaries, and your interpretations. Distinguish event time, note time, and follow-up time. For every action, show actor, requested, promised, completed, evidence, owner, and status; never infer completion. Flag contradictions and duplicates without overwriting prior history. Stop on identity, chronology, taxonomy, or status uncertainty. Do not write to the CRM, close the case, change a disposition, or contact the customer; I verify and enter the record.", "humanReviewPoint": "The representative verifies identity and content, chooses classifications and disposition, confirms action status, and enters the approved record.", "commonFailure": "Compressing a conversation into a smooth summary that changes who said what, scrambles the timeline, or converts a requested or promised action into a completed one." }, { "id": "record-correction-amendment", "goal": "Correct a material interaction-record error through an attributable amendment without erasing the original history.", "inputs": [ "Existing record and immutable record ID", "Source evidence supporting the correction", "Correction reason, requester, discovery time, and applicable retention rule", "Current open actions and downstream case decisions affected" ], "connectedSystems": [ "Read-only existing record", "Approved source-evidence repository", "Draft amendment surface with no overwrite permission" ], "steps": [ "Identify the exact field or statement at issue, the original author and timestamp, and how the error was discovered.", "Compare the original record to the source evidence and distinguish transcription error, later-acquired fact, disputed account, taxonomy change, and action-status update.", "Draft an amendment that states the old value, proposed value, reason, source, author, and amendment time; preserve the original text and record ID.", "Map downstream effects on promises, due dates, escalation, reporting, privacy disclosure, billing, or case disposition.", "List the authorized record action and any notification required by policy, but do not perform either.", "Stop for the designated record owner to approve and append the amendment; never backdate, silently overwrite, or delete history." ], "expectedArtifact": "Attributable correction packet with evidence comparison, old-to-new diff, downstream-impact map, and proposed append-only amendment", "starterPrompt": "Prepare an append-only correction packet for this interaction record using the existing record, source evidence, correction request, and retention rule. Identify the exact disputed field, original author and timestamp, and whether this is a transcription error, later fact, disputed account, taxonomy change, or status update. Show old value, proposed value, reason, source, amendment author and time, and downstream effects. Preserve the original record and ID. Do not backdate, overwrite, delete, notify, or change the CRM; the designated record owner approves and appends the correction.", "humanReviewPoint": "The designated record owner decides whether evidence supports correction, approves downstream handling, and appends the amendment under retention policy.", "commonFailure": "Improving the current record by silently replacing the old account, destroying who-knew-what-when provenance and obscuring decisions made from the original information." } ], "setupRecommendations": [ { "item": "Provide authorized transcripts or contemporaneous notes, the current record schema and taxonomy, read-only case context, retention rules, and a draft surface with no CRM write or outbound-send permission.", "rationale": "Reliable interaction records require provenance, chronology, and correction history; direct writes would let a summary error become durable customer history.", "capabilityIds": [ "connected-apps-files", "approval-enterprise-controls" ] } ], "prompts": [ { "id": "build-attributed-chronology", "title": "Reconstruct who said and did what, when", "prompt": "From this authorized transcript or note, build an event chronology that separates event time, statement time, note time, and proposed follow-up time. Attribute each item to customer, representative, system record, or inference. Preserve material quotations exactly and flag inaudible, ambiguous, or contradictory passages. Stop if identity or sequence cannot be established. Return the chronology only; I verify it before record entry.", "expectedArtifact": "Attributed interaction chronology with time types and ambiguities", "humanReviewPoint": "Verify identity, sequence, quotations, and time interpretation." }, { "id": "populate-record-fields", "title": "Draft the record without laundering interpretation into fact", "prompt": "Populate this approved interaction-record schema from the attributed chronology and read-only case context. Keep verbatim quotes, factual system evidence, summaries, and interpretations in separate labeled fields. Use only supplied taxonomy values and mark REQUIRED-UNKNOWN where evidence is absent. Return the record draft and missing-field list. Do not guess a code or write to the CRM; I choose classification and approve entry.", "expectedArtifact": "Schema-complete record draft with labeled evidence and missing fields", "humanReviewPoint": "Choose classification and disposition, resolve missing fields, and enter the approved record." }, { "id": "audit-action-status", "title": "Distinguish requested, promised, and completed actions", "prompt": "Audit this interaction record and case history for actions. For each, show actor, requested action, promised action, completed action, completion evidence, owner, due condition, current status, and source. Flag any status inferred from intent, an unsourced completion, or a promise without an owner. Return an action-status ledger and stop. Do not mark complete, close the case, or contact anyone; I verify status and assign follow-up.", "expectedArtifact": "Evidence-backed action-status ledger", "humanReviewPoint": "Confirm completion evidence, assign ownership, set disposition, and authorize follow-up." }, { "id": "draft-append-only-correction", "title": "Correct the record without erasing history", "prompt": "Using this existing record, correction request, source evidence, and retention rule, draft an append-only amendment. Show the original field and timestamp, proposed value, correction type, reason, source, author, amendment time, and downstream effects. Preserve the original record and ID. Stop if evidence is disputed or authority is unclear. Do not backdate, overwrite, delete, or update the CRM; the record owner decides and appends the correction.", "expectedArtifact": "Correction packet with old-to-new diff and provenance", "humanReviewPoint": "The record owner approves the evidence, amendment, and downstream notifications." } ], "humanOwnedResponsibilities": [ "Record-entry law: ChatGPT prepares outside the customer system; an authorized human verifies and enters every interaction record and disposition", "Attribution law: customer words, representative statements, system facts, summaries, and model interpretations remain separately labeled", "Chronology authority: a human resolves ambiguous event time, note time, sequence, identity, and contradictory source evidence", "Action-status law: requested, promised, assigned, and completed are distinct states; only source evidence and the responsible human establish completion", "Correction law: material changes are attributable, append-only amendments; no silent overwrite, deletion, or backdating", "Follow-up authority: case closure, escalation, ownership, due dates, and customer contact remain with the designated representative or supervisor" ], "risks": [ { "id": "summary-distortion", "level": "high", "description": "A concise record can change the speaker, omit a qualification, or turn an interpretation into a customer statement or system fact.", "mitigation": "Preserve material quotations, label every evidence class, retain source locators, and require human comparison to the transcript before entry." }, { "id": "false-action-completion", "level": "high", "description": "A requested, intended, or promised action can be recorded as completed, suppressing follow-up and creating a false service history.", "mitigation": "Require an action-status ledger with actor, state, completion evidence, owner, and source; only a human confirms completion and closure." }, { "id": "history-erasure", "level": "high", "description": "Silently correcting an earlier note can destroy provenance, hide what prior decisions relied on, and violate retention or audit expectations.", "mitigation": "Use append-only amendments with old-to-new diffs, source evidence, author and time, downstream-impact review, and record-owner approval." } ], "sources": [ { "id": "occupation-customer-service-representatives", "title": "Customer Service Representatives occupational render model", "url": "https://singulariki.com/roles/customer-service-representatives", "publisher": "gold.onet_role_starter_surface", "lastReviewedAt": "2026-06-03" }, { "id": "openai-chatgpt-work", "title": "ChatGPT Work and Codex", "url": "https://help.openai.com/en/articles/20001275-chatgpt-work-and-codex", "publisher": "OpenAI", "lastReviewedAt": "2026-07-11" }, { "id": "openai-work-deliverables", "title": "Creating and editing documents, spreadsheets, and presentations with ChatGPT Work", "url": "https://help.openai.com/en/articles/20001278-creating-and-editing-documents-spreadsheets-and-presentations-with-chatgpt-work", "publisher": "OpenAI", "lastReviewedAt": "2026-07-11" }, { "id": "openai-work-cloud-browser", "title": "Using cloud browser in ChatGPT", "url": "https://help.openai.com/en/articles/20001280-using-cloud-browser-in-chatgpt", "publisher": "OpenAI", "lastReviewedAt": "2026-07-11" }, { "id": "openai-work-desktop-browser", "title": "Using the built-in browser in the ChatGPT desktop app", "url": "https://help.openai.com/en/articles/20001277-using-the-built-in-browser-in-the-chatgpt-desktop-app", "publisher": "OpenAI", "lastReviewedAt": "2026-07-11" }, { "id": "openai-chatgpt-skills", "title": "Skills in ChatGPT", "url": "https://help.openai.com/en/articles/20001066", "publisher": "OpenAI", "lastReviewedAt": "2026-07-11" }, { "id": "openai-chatgpt-scheduled-tasks", "title": "Scheduled Tasks in ChatGPT", "url": "https://help.openai.com/en/articles/10291617-tasks-in-chatgpt", "publisher": "OpenAI", "lastReviewedAt": "2026-07-11" }, { "id": "openai-workspace-agents", "title": "ChatGPT Workspace Agents for Enterprise and Business", "url": "https://help.openai.com/en/articles/20001143", "publisher": "OpenAI", "lastReviewedAt": "2026-07-11" } ], "freshness": { "platformVersion": "2026-07-09-launch", "platformReviewedAt": "2026-07-11", "occupationalDataVersion": "O*NET 30.3", "occupationalDataGeneratedAt": "2026-06-03T05:05:52.989984+00:00" }, "quality": { "uniqueOccupationalOrTaskEvidence": false, "validPlatformSpecificWorkflow": true, "clearOutputArtifact": true, "realCapabilityMatch": true, "meaningfulHumanReviewSection": true, "sourcesAndFreshness": false, "distinctInternalRelationships": true, "noUnresolvedFactualOrProductVersionFailure": false, "indexable": false } }