{ "platformId": "chatgpt-work", "taskSlug": "task-18565", "directAnswer": "Use ChatGPT Work to turn a customer conversation into a review-ready service packet: preserve what the customer actually said, identify the request and promised follow-up, check product and policy claims against approved sources, draft the reply, and show exactly which order, cancellation, refund, or account action would still require your approval. Keep it on the preparation side of the counter. A fluent draft is not a company commitment: ChatGPT must not quote an unapproved price, promise an outcome, enter an order, cancel an account, issue value, or contact the customer. The usage figures attached to this task are projected role-task priors, not measured ChatGPT adoption by customer service representatives, so they support a controlled trial rather than a claim about current practice.", "topUseCases": [ { "interventionId": "task-18565:customer-interaction-to-reviewed-draft:approval-required", "taskId": "task-18565", "task": "Confer with customers by telephone or in person to provide information about products or services, take or enter orders, cancel accounts, or obtain details of complaints.", "workflowId": "customer-interaction-to-reviewed-draft", "authorityMode": "approval-required", "centralityScreen": { "value": 0.47724569899999997, "importance": { "value": 0.9175, "sourceField": "role-tasks-kv:task-18565.importance" }, "occupationalRelevance": { "value": 0.943, "sourceField": "role-tasks-kv:task-18565.relevance / 100" }, "frequency": { "kind": "modal-response-share", "modalResponseSharePct": 55.16, "collectionDate": "2026-06-03", "sourceField": "role-tasks-kv:task-18565.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-18565].observed-distribution" }, "evidenceConfidence": "weak", "withinGateScore": 0.38764281901275, "gate": { "status": "approval-required", "reason": "A person must approve the verifiable handoff before action." }, "capabilityIds": [ "connected-apps-files", "deliverable-generation", "approval-enterprise-controls" ], "expectedArtifact": "Customer service packet with a verbatim fact record, source-linked answer, commitment ledger, and unsent response draft", "humanReviewPoint": "Verify the customer record and every policy or product claim, decide the remedy or transaction, and personally perform any order, cancellation, account, value, or send action.", "riskLevel": "high" } ], "workflows": [ { "id": "customer-interaction-to-reviewed-draft", "goal": "Turn one customer conversation into an evidence-separated record and an unsent answer without creating a commercial or account commitment.", "inputs": [ "Authorized transcript or contemporaneous notes with channel and interaction time", "Current product catalog, terms, service policy, and escalation matrix", "Read-only customer and order context needed for this inquiry", "Approved voice guidance and required record fields" ], "connectedSystems": [ "Approved product and policy sources", "Read-only customer context", "Draft surface with no send or transaction permissions" ], "steps": [ "Capture the customer's words, identifiers, request, and stated impact; preserve material quotations verbatim and label summaries as summaries.", "Build a claim ledger separating customer assertions, company-record facts, policy text, representative observations, and unresolved questions; attach a source and effective date to each company claim.", "Identify the requested outcome and every implied commitment, including price, availability, delivery, eligibility, refund, cancellation, credit, or follow-up timing; route anything outside delegated authority to the named approver.", "Draft a response that answers only source-supported points, marks NEEDS-SOURCE or NEEDS-DECISION inline, and does not turn empathy language into an admission or promise.", "Return the interaction record, claim ledger, proposed next action, and unsent reply with a change receipt naming any order, account, value, or communication action a human would perform.", "Stop if identity, policy version, transaction scope, or approval authority is uncertain; do not contact the customer or change any system." ], "expectedArtifact": "Review-ready customer service packet containing the interaction record, claim ledger, commitment ledger, proposed disposition, and unsent reply", "starterPrompt": "Prepare a customer service review packet from this authorized transcript, read-only account context, current product sources, and policy library. Preserve material customer quotations verbatim. Separate customer assertions, company-record facts, policy text, and your inferences; cite the source and effective date for every product or policy claim. List every implied promise or transaction and mark who may approve it. Draft an unsent reply using NEEDS-SOURCE and NEEDS-DECISION where evidence or authority is missing. Stop if identity, policy version, eligibility, price, availability, or approval authority is unclear. Do not enter an order, cancel an account, issue value, promise an outcome, or contact the customer; I decide the disposition and perform every external action.", "humanReviewPoint": "The authorized representative verifies identity and sources, chooses the disposition, approves all commitments, enters any transaction, and sends the response.", "commonFailure": "Converting a customer's account of events or an outdated policy snippet into a confident company promise, then allowing the polished reply to outrun transaction authority." }, { "id": "complaint-to-escalation-brief", "goal": "Turn a complaint with disputed facts or requested value into a bounded escalation brief without prejudging fault or remedy.", "inputs": [ "Customer's complaint and requested resolution", "Interaction and transaction history available to this representative", "Applicable complaint, refund, safety, privacy, and escalation policies", "Delegation limits and named escalation owners" ], "connectedSystems": [ "Approved case-history view", "Current policy and escalation library", "Draft escalation queue with no customer-send permission" ], "steps": [ "Create an event timeline that attributes each statement to the customer, company record, employee note, or system event.", "List agreements, contradictions, missing evidence, and records that must be preserved; do not resolve a contradiction by choosing the most fluent narrative.", "Screen for mandatory escalation triggers such as safety, privacy, discrimination, legal threat, chargeback, vulnerable customer, or repeated service failure using the supplied policy only.", "Compare permitted resolution paths without assigning fault, calculate any proposed value from the approved rule, and identify the approver for each path.", "Draft an internal escalation brief and a minimum-commitment acknowledgment for review; include a stop list of claims and remedies that cannot yet be stated.", "Stop before assigning fault, offering value, closing the complaint, or sending either draft." ], "expectedArtifact": "Complaint escalation packet with attributed timeline, evidence gaps, trigger screen, remedy options, and unsent acknowledgment", "starterPrompt": "Build an escalation packet for this complaint using only the interaction history, transaction evidence, current policy, and authority matrix I provide. Attribute every timeline entry to its source. Separate confirmed events, customer allegations, employee notes, contradictions, and missing evidence. Screen the supplied escalation triggers and show the approver for every remedy option. Draft an internal brief and an unsent acknowledgment that makes no admission or unsupported promise. Stop if identity, policy scope, material facts, or mandatory escalation status is unresolved. Do not decide fault, issue value, close the case, or contact the customer; those actions remain with the authorized representative.", "humanReviewPoint": "A human case owner validates the record, decides escalation and remedy, approves any admission or commitment, and controls all customer contact.", "commonFailure": "Collapsing disputed accounts into a single authoritative story or using a generic apology that inadvertently admits liability or promises a remedy." } ], "setupRecommendations": [ { "item": "Connect only the approved product catalog, current policy library, and read-only customer context; keep order entry, account changes, credits, cancellations, and outbound channels outside the workspace.", "rationale": "The useful delegation is evidence retrieval and draft preparation. Transaction or send access would let an unsupported inference become a customer-visible commitment.", "capabilityIds": [ "connected-apps-files", "approval-enterprise-controls" ] } ], "prompts": [ { "id": "build-customer-fact-record", "title": "Separate the customer's account from company facts", "prompt": "From this authorized transcript and read-only account history, build a fact record with four columns: verbatim customer statement, company-record fact with source and timestamp, representative observation, and unresolved question. Preserve material quotations exactly and do not infer identity, motive, fault, or eligibility. Return the record plus the smallest questions needed to continue. Stop if the customer or account cannot be reliably identified. Do not update the case; I verify and approve the record.", "expectedArtifact": "Attributed customer fact record with source timestamps and unresolved questions", "humanReviewPoint": "Verify identity, quotations, and record sources before approving the case note." }, { "id": "check-product-policy-claims", "title": "Check every answer against current product and policy sources", "prompt": "Audit this proposed customer answer against the approved product catalog and policy set I provide. For every claim about price, availability, terms, delivery, eligibility, cancellation, refund, credit, or timing, cite the controlling source and effective date or mark NEEDS-SOURCE. Identify conflicts between sources and stop rather than choosing one. Return a claim ledger and corrected draft; do not send it or make a commitment. I resolve source conflicts and approve the answer.", "expectedArtifact": "Source-linked claim ledger and corrected unsent answer", "humanReviewPoint": "Resolve conflicting or stale sources and approve every customer-facing claim." }, { "id": "map-customer-commitments", "title": "Expose the commitments hidden in a routine reply", "prompt": "Review this interaction and draft reply for explicit or implied commitments. List every promise, price, inventory statement, delivery date, remedy, refund, credit, cancellation, account change, and follow-up deadline; show its evidence, delegated authority, and required approver. Rewrite any unauthorized commitment as a bounded next step. Stop if authority is absent. Do not transact or contact the customer; I choose and execute the disposition.", "expectedArtifact": "Commitment ledger with authority status and bounded rewrite", "humanReviewPoint": "Approve the commercial or account decision and perform every transaction." }, { "id": "prepare-complaint-escalation", "title": "Prepare a complaint escalation without deciding the case", "prompt": "Using this complaint, interaction history, current escalation policy, and authority matrix, create an attributed event timeline, contradiction list, missing-evidence list, mandatory-trigger screen, and remedy options by approver. Draft an internal escalation brief and an unsent minimum-commitment acknowledgment. Do not assign fault, admit liability, offer value, close the complaint, or send anything. Stop on unresolved identity, safety, privacy, legal, or mandatory-escalation questions; the human case owner decides the path.", "expectedArtifact": "Evidence-separated complaint escalation packet", "humanReviewPoint": "The case owner decides fault, escalation, remedy, closure, and customer communication." } ], "humanOwnedResponsibilities": [ "Commitment law: only an authorized employee may bind the company on price, availability, delivery, eligibility, remedy, refund, credit, cancellation, or follow-up timing", "Transaction law: order entry, account change, cancellation, refund, credit, case closure, and outbound contact remain human actions even when the draft is routine", "Evidence separation: customer assertions, company records, policy text, employee observations, and model inferences remain visibly distinct", "Identity and disclosure judgment: a human verifies the customer and decides what account, order, and complaint information may be discussed", "Complaint judgment: fault, admission, mandatory escalation, remedy, and closure stay with the named case owner", "Source conflict authority: a person resolves stale or contradictory product and policy sources before any claim reaches the customer" ], "risks": [ { "id": "unsupported-customer-commitment", "level": "high", "description": "A plausible response can turn an unverified product fact, eligibility inference, or empathy phrase into a promise the company did not authorize.", "mitigation": "Require a source-and-effective-date claim ledger, a separate commitment ledger, and human approval of every price, timing, remedy, or account consequence before sending." }, { "id": "disputed-fact-collapse", "level": "high", "description": "Customer allegations, employee notes, and system events can be blended into one confident narrative, obscuring contradictions and prejudging fault.", "mitigation": "Use an attributed event timeline, preserve material quotations, label every inference, and stop for a case owner when evidence conflicts." }, { "id": "unauthorized-account-action", "level": "high", "description": "A correct-looking draft can be mistaken for authority to enter an order, cancel service, issue value, close a complaint, or disclose account information.", "mitigation": "Keep transaction and send systems disconnected, include a human-action receipt in every packet, and require verified identity plus named 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 } }