Hats · Support
AI agent for support teams
Give your support team an AI agent that knows your help center, your macros and your escalation paths.
The Support Hat triages tickets, drafts replies and summarizes issues for engineering, so agents clear the queue faster without going off-script.
Hey skilder, draft a reply to Initech’s refund request and send it
Support Hat loaded
Reply drafting
Drafted a reply citing your 30-day refund policy and next steps. Send it to Initech?
Yes, send it
Summarize the login bug thread for engineering
Escalation summary
4 users, SSO timeout after 30s, repro steps + logs attached. Filing it for engineering.
Inside the Support Hat
A Hat bundles everything the agent needs to do a role, its mission, your rules, the skills it can run, and the tools it connects to. Building blocks you assemble.
Role description
I triage tickets, draft on-brand replies and escalate cleanly to engineering, using your help center, macros and past resolutions so customers get consistent answers.
Instructions
"When running the Reply drafting skill: match the customer’s tone, cite the help-center article, and never promise a refund above policy without approval."
Skills
Hover a skill to see what it does.- Ticket triage Classifies and prioritizes incoming tickets, routing each to the right queue.
- Reply drafting Drafts an on-brand reply grounded in your help center and past resolutions.
- KB lookup Finds the right help-center article or macro for the issue.
- Escalation summary Summarizes a thread for engineering with steps to reproduce.
- CSAT follow-up Drafts a follow-up after resolution to close the loop.
Connectors
- Zendesk
- Intercom
- Jira
- Confluence
- Slack
How it works
Live in minutes. Three simple steps.
- 1
Pick a Hat
Start from a Hat and shape it to your playbook, prices, rules, tone. No code, minutes not weeks.
- 2
Plug in your tools
Connect Salesforce, HubSpot and the rest, so the agent works on your real data, not generic answers.
- 3
Run it from your AI
Your team wears the Hat inside Claude, ChatGPT or Copilot, governed and audited, no new tool to learn.
Frequently asked questions
-
Does it answer using our help center and policies?
Yes. The Support Hat grounds every reply in your help-center articles, macros and refund/escalation policies, so answers stay consistent and on-policy, not improvised.
-
Which helpdesk and tracking tools does it connect to?
It connects to Zendesk, Intercom, Jira and Confluence, so it reads tickets, drafts replies and files clean escalations.
-
Can agents use it inside their existing tools?
Yes. The Support Hat is LLM-agnostic: it runs in Claude, ChatGPT, Microsoft Copilot or any AI assistant, no new helpdesk to migrate to.
-
Is customer data kept private?
Yes. skilder enforces workspace isolation and full audit logging of every run; customer data stays in your tenant and is never used to train public models.
From scattered prompts to governed, shared roles.
Try skilder (no credit card), or scope your use case with our founders.
