Support and operations

Triage, draft, and file repeat ops work

Carina reads exports you provide, groups themes, writes drafts, and saves outputs to disk. It is not a hosted helpdesk; it runs on your infrastructure with your credentials.

Who it is for

Small support teams, founders handling inbox load, and ops staff who want an agent to process batches instead of copying from a chat window.

Problem it solves

Weekly ticket summaries, reply templates, and internal status notes take the same steps every time: read, cluster, write, save.

Outputs you get

  • Ticket theme summaries and priority groupings
  • Draft reply templates and internal notes as files
  • Optional outbound email when email-send is configured and confirmed
  • Session memory so follow-up tasks reuse prior context

Config and integrations

  • Core tools: file-read, file-write, web-search, email-send (when API keys set)
  • Gateways: CLI, web UI, Telegram, WhatsApp, Slack (each needs credentials in ~/.carina/.env)
  • Optional pack: productivity for Google, Notion, Airtable integrations
  • PostgreSQL + Redis for memory and sessions on self-hosted installs

Example workflow

  1. You export last week's support tickets to a CSV in ./inbox/.
  2. Carina reads the file, groups recurring issues, and writes a summary plus a reply template.
  3. Outputs land in ./outbox/ for human review before anything is sent.

"Summarise ./inbox/tickets-last-week.csv by theme and save a reply template to ./outbox/reply-template.md"

Install Carina Channel setup

For production deployments with audit requirements, pair Scout so tool calls are logged.