Intercom Prompts

Design a Fin AI Copilot workflow for [COMPANY] support agents that drafts accurate replies from our Knowledge Hub articles before an agent hits send. Include: when to accept, edit, or override Copilot drafts; which [TICKET_TYPE] tickets should stay human-only; and a 5-step review checklist agents follow inside the Inbox. Output as a numbered SOP our team can pin in the help center and train on this week.

Audit the [COMPANY] Knowledge Hub to find content gaps that hurt Fin AI Agent resolution rates. Review our top 20 recurring conversation topics, flag missing or outdated articles, and recommend: which 10 articles to write first, which existing articles to rewrite, and which sources (help center, internal docs, macros, PDFs) Fin should trust most. Deliver a prioritized table with owner and deadline columns for [TEAM_NAME].

Create a deflection content pack for [COMPANY] so Fin AI Agent resolves more conversations without human handoff. Write 8 short, plain-language answers for our most repeated questions about [PRODUCT_AREA], each under 120 words, with a clear next step or link. Then list the 5 questions Fin should always escalate, plus the exact escalation message and Inbox assignment rule to use. Format it for direct import into the Knowledge Hub.

Build a set of Intercom Workflow routing rules for [COMPANY] that triage incoming conversations automatically. Define rules by channel (chat, email, WhatsApp), intent ([INTENT_1], [INTENT_2], [INTENT_3]), and customer tier, then map each to the right outcome: Fin resolution, [TEAM_A], [TEAM_B], or priority escalation. Include fallback logic for unrecognized intents and a test plan with 15 sample conversations before going live.

Design a custom AI Scorecard for [COMPANY] to QA 200 support conversations per month in Intercom. Define 6 scored criteria (greeting, accuracy, tone, resolution, follow-up, brand voice) with clear pass/fail rubrics and weights. Specify which conversation types to sample, how to handle Fin-resolved vs human-resolved threads, and a monthly calibration process for [QA_LEAD]. Output the scorecard as a table plus a one-page reviewer guide.

Write a CSAT recovery playbook for [COMPANY] triggered when a conversation gets a low rating in Intercom. Cover: an immediate outbound follow-up message template, who owns the recovery (agent vs [MANAGER_ROLE]), when to reopen the ticket, and how to log the root cause for reporting. Include 3 ready-to-send apology templates for common failure types in [SUPPORT_CHANNEL] and a weekly review ritual to prevent repeats.

Build a starter macro library for [COMPANY] agents in the Intercom Inbox. Create 12 reusable macros for [PRODUCT_AREA]: greetings, status updates, troubleshooting steps, billing answers, escalation handoffs, and closing messages. Each macro needs placeholders for [CUSTOMER_NAME] and [TICKET_ID], a short internal title, and guidance on when to use it vs Fin automation. Deliver them as a copy-paste list with usage notes.

Define a weekly support analytics report for [COMPANY] using Intercom reporting data. Specify the 10 metrics to track (Fin resolution rate, conversation volume, first response time, resolution time, CSAT, top topics via AI Topics, backlog by [TEAM_NAME], Copilot adoption, deflection rate, escalation rate), where each metric lives in the dashboard, and red/yellow/green thresholds. Include a one-paragraph summary template for [LEADER_NAME].

Create a chatbot training intent map for [COMPANY] to teach Fin AI Agent our top customer intents. List 15 intents for [PRODUCT_AREA] with example phrasings customers actually use, the correct answer source, any required action (refund, booking change, account lookup), and escalation triggers. Mark which intents Fin can fully resolve vs assist-only, and note which integrations Fin may call to complete each task.

Design a 2-week onboarding sprint for new [COMPANY] support agents using Fin AI Copilot as their training partner. Week 1: product knowledge, Inbox basics, and shadowing Copilot drafts with a buddy agent. Week 2: supervised live conversations, macro practice, and a graded QA review against our scorecard. Include daily checklists, a final sign-off test for [MANAGER_ROLE], and success criteria for handling [TICKET_TYPE] solo.