Sync Intercom help center with product release notes using AI
Build a workflow that helps a documentation team keep their Intercom Help Center in sync with product "What's New" announcements. Whenever the product ships an update, the workflow figures out which existing help articles should change (or whether a brand-new article is needed), drafts the update text, and says exactly where it fits. Treat the Help Center as read-only — never write back to it; every draft is a suggestion a human copies in. Trigger it two ways. First, on a schedule (a cron every few hours, also runnable on demand): scrape the most recent posts from the product's public "What's New" page — or wherever your release notes live — including each post's body, and fetch all published Help Center articles from Intercom — or your docs/knowledge-base platform. Cache both in a persistent volume so the other steps share the same data. Second, expose an internal review page (a route a person opens in the browser) that lists the latest updates and lets someone pick one, or paste any post URL, to analyze on demand. The core is an AI agent (use Claude via Anthropic — or your preferred model) that scans the article index, reads candidate articles, and returns structured suggestions: which articles to update, why, where the edit goes, and a copy-ready draft — proposing a new article when nothing fits. Wire this into two places: the review page (analyze a chosen post interactively and show the drafted suggestions) and an automatic path that runs after each scheduled refresh, detecting genuinely new posts (seed a baseline on first run so activation doesn't create a backlog burst) and drafting for each. For new posts, open a ticket/case capturing the suggestions — one block per suggested edit or new article, laid out as article-to-update → why → where it fits → drafted update. Use Tines Cases — or your team's ticketing/case tool. Also let the review page create a case on demand from whatever suggestions it's showing. After cases are created, post a notification to a team channel — use Slack — or your team's chat tool — summarizing what was opened, and send nothing if there was nothing new. Keep all internal routes authenticated to space members. Make the source feed, post count, relevance judging, and draft style easy to tweak in one place. Tools used: Tines What's New page (web scraping), Intercom Articles API, Anthropic (Claude) API, Tines Cases API, Slack Web API (chat.postMessage)
What this prompt builds
This workflow automatically synchronizes an Intercom Help Center with product release notes by identifying which articles need updates when new features ship. An AI agent analyzes the latest "What's New" posts, matches them to existing help articles, drafts update suggestions with exact placement guidance, and creates tracking cases for documentation teams. The workflow runs on a schedule and provides an on-demand review interface for manual analysis.
The problem
Documentation teams struggle to keep help center articles current when new product features ship. Manually reviewing every release note against hundreds of existing articles to identify what needs updating is time-consuming and error-prone, and teams often lose track of which updates are still pending. This workflow solves that by automatically analyzing new product "What's New" announcements, matching them to relevant Intercom help articles, and drafting the specific update text with placement guidance. It creates tracking cases for each suggested change so nothing falls through the cracks, turning a tedious audit process into a queue of ready-to-review drafts.
Solution and impact
Documentation teams receive AI-drafted article updates with exact placement instructions whenever a new product feature ships, eliminating the manual work of scanning release notes against the entire help center. The workflow creates a case for each batch of suggestions and notifies the team in Slack, giving them a clear queue to review and apply rather than a vague backlog. By running on a schedule and offering an on-demand review interface, it ensures the help center stays synchronized with product changes without requiring anyone to remember which articles are out of date.