Tickets and metrics
| Concept | Description | Example |
|---|---|---|
| Support ticket | A record of a customer’s request for support, created in your ticketing system. | ”The ticket is about a technical issue.” |
| Assigned tickets | Tickets assigned to your Customer Agent in your ticketing system. | ”We routed 30 assigned tickets to the agent this week.” |
| Engaged ticket | A ticket where your Customer Agent performed one or more engage actions: replying to the customer, auto-closing spam, or merging duplicate tickets. A ticket the agent engaged with but did not resolve was previously called an “assisted ticket." | "The agent engaged 40 tickets this week.” |
| Resolved ticket | A ticket your Customer Agent engaged with, then closed, that stays closed for at least 3 days. | ”The ticket was resolved by our Customer Agent.” |
| Engagement rate | Tickets your Customer Agent engaged with, divided by total tickets created in a period. | ”Our engagement rate is up 7 points.” |
| Resolution rate (of engaged) | Tickets your Customer Agent resolved, divided by tickets it engaged with in a period. | ”Our resolution rate of engaged tickets is 83%.” |
| Resolution rate (of assigned) | Tickets your Customer Agent resolved, divided by tickets assigned to it in a period. | ”Our resolution rate of assigned tickets is at an all-time high.” |
| Resolution rate (of all) | Tickets your Customer Agent resolved, divided by all tickets created in a period. | ”Our resolution rate of all tickets crossed 50%.” |
| Customer sentiment score | The share of tickets where customer sentiment is positive or neutral. | ”The sentiment score is 70%.” |
Agents
| Concept | Description | Example |
|---|---|---|
| Customer Agent | The AI agent that resolves your customers’ support tickets end to end, working inside your ticketing system. | ”Our Customer Agent resolved 200 tickets this week.” |
| Internal Agent | The AI agent you prompt in natural language to build, test, and improve your Customer Agent and everything around it - also referred to as just “Parahelp.” Nothing it proposes goes live without your approval. | ”Ask the Internal Agent to draft a refund tool.” |
| Customer Agent configurations | Everything that shapes the agent’s behavior: • Channels: the channels it handles tickets through • Base tools: built-in actions always available to it, managed by Parahelp • Custom tools: actions built by the Internal Agent, like refunds or account lookups • Persistent memory: instructions it follows on every ticket, like personality instructions • Working memory: instructions it follows only for related tickets, like refund requests or billing issues instructions • Knowledge snippets: internal knowledge it can search • Web pages: your scraped web pages, available for it to search | ”Have you updated our configurations?” |
| Flow | The behavior your team builds into the Customer Agent for a ticket type. A flow combines policies, knowledge, and tools - including the tool connections its actions run through. | ”We shipped a new refund flow last week.” |
| Transfer | When the Customer Agent hands a ticket to a human. | ”The agent transferred the ticket to our billing team.” |
| Release | A published set of Customer Agent configuration changes. | ”Publish your release so the team stays in sync.” |
| Task | When you prompt the Internal Agent, it opens a chat you keep prompting until the task is done. | ”I’ll start a task to update our configs.” |
| Action | An action the Internal Agent proposes in a task (a config update, a Linear issue, a Notion page). Needs approval by default, and you can auto-approve selected actions per task or automation. | ”I approved the action to update the configs.” |
| Automation | An Internal Agent task that runs on a schedule or event, from a set prompt. | ”Can you set up an automation for this?” |
| Test simulation | A simulation of how your Customer Agent handles a scenario. Includes a summary, the steps, and an evaluation: failed, neutral, or passed. | ”It ran 5 test simulations that all passed.” |
| Todo | Work the Internal Agent identified and assigned to a user or team, ready to start with one click. It creates todos automatically for high-leverage improvements it finds in automations and one-off tasks. | ”I started the todo it created for the refund gap.” |
Channels and connections
| Concept | Description | Example |
|---|---|---|
| Channel | The channels the Customer Agent handles tickets through: email, live chat, and Slack via your ticketing system, or the API. | ”We added live chat as a channel.” |
| Connection | A link between Parahelp and one of your tools, like your ticketing system, Linear, or Slack. • Customer Agent: uses scoped tools to take actions. Tools leverage the connections, like a tool that executes a refund using the Stripe connection. • Internal Agent: uses connections as it codes to complete tasks and build the scoped tools for the Customer Agent. | ”Set up the Slack connection so the agent can post reports.” |