Draft a support reply from a ticket and its guidance
A useful support draft starts from the customer’s exact ticket and the team guidance that applies. The draft stays in the AI client until a person decides what to do with it.
For support teams using a compatible AI client with Zendesk and Notion connected.
What needs to be in place
- In My apps, add Zendesk and Notion, then connect the accounts you want your AI client to use.
- The Zendesk ticket ID and the title or distinctive phrase from the relevant Notion article.
In My apps, add the apps listed above and connect the accounts you want your AI client to use. Then connect your AI client to Weavz. A workspace template is optional: it can configure app access and defaults, but your AI client runs the task.
Work through the selected sources
The exact actions available depend on the integrations and actions enabled in the workspace. These steps keep the task bounded to the account and records you selected.
Read the exact ticket
Confirm the ticket requester and issue. Include the comment history only if the current message does not contain enough context.
Find and read the guidance
Search for the support article by its known title, confirm the page ID and title, then read its actual body. Start at one child-block level and expand only the blocks needed. If several pages match, choose the correct policy or ask the teammate.
Prepare a reply
Draft a concise answer in the team’s tone. Quote or link to the source guidance and mark any claim that the records do not support.
App actions and developer details
Action names and inputs
These details help when configuring or implementing the task. You can follow the guide without starting from the action identifiers.
zendesk.get_ticketRead a specified ticket; request comments only when they are needed.
Required inputs:
ticket_idnotion.find_pageFind the internal guidance page by title and return its page ID; this search does not return the page body.
Required inputs:
titlenotion.getPageOrBlockChildrenRead the selected page body as block data at depth 1, then expand only child blocks needed for the reply. Set parentId to the selected page or block ID, markdown=false, and dynamic.depth=1.
Required inputs:
parentIdmarkdowndynamic
A starting instruction for your AI client
Replace bracketed details with the records and scope you chose. Review the returned result against its sources.
Example instruction
Review Zendesk ticket [ticket ID] and its relevant comments if the ticket message needs more context. Search Notion for the support article [article title], confirm the exact page, then read its body. Start with one child-block level and expand only the blocks needed to answer the ticket. Draft a reply using only facts supported by the ticket and page content. Include links or IDs for both sources and list missing information or policy questions. Keep it as a draft in this response. Do not add a Zendesk comment, send email, or change the ticket.
What the result contains
A customer-ready draft plus source references and a short list of unresolved facts for the support teammate.
Keep the source trail with it
Carry the ticket ID and Notion page link into the review. The Notion search result only identifies the page; compare the draft with the page body returned by the separate block-content read.
Keep drafts and writes separate
This task reads Zendesk and Notion, then returns a draft. Publishing a public comment is a separate Zendesk write action and should not be inferred from a request to draft.
If a source or connection is missing
If ticket comments were omitted, request them only when necessary. If no relevant Notion article is found, or the selected page has no readable body, stop and ask which guidance applies instead of inventing policy. If the article has nested blocks, expand only the relevant children and preserve their source links.
Set up the accounts and task
Optional workspace template: Support Ops Center.
Integrations
The template includes Zendesk and Notion for ticket context and support knowledge.