AI Ticketing
AI Ticketing adds optional, opt-in AI assistance to your help desk. When you turn it on, Ascent can:
- Auto-categorize new tickets — as each ticket is created, Ascent suggests the best-matching category and a priority, with a confidence score and a one-line reason. A technician can accept the suggestion (which applies it to the ticket) or dismiss it.
- Suggest replies — when working a ticket, a technician can generate a draft reply based on the ticket and its conversation, then edit it before sending.
You choose which AI provider powers it: Anthropic (Claude), OpenAI, OpenRouter, or a local / custom OpenAI-compatible endpoint you host yourself (for example Ollama or LM Studio). See Choose a provider.
AI Ticketing is off until you turn it on, and it is only available on plans that include it. While it is off, no AI options appear and no ticket content is ever sent to any provider. Nothing leaves Ascent unless you both enable the feature and configure a usable provider.
Availability
AI Ticketing must be included in your plan. If your organization is not licensed for it, the AI Ticketing settings page and the on-ticket AI controls do not appear. Contact your account owner about upgrading if you do not see it.
Who can configure it
Configuring AI Ticketing requires the Organization permission (Owners and Admins by default). The settings page lives under Administration → AI Ticketing (/settings/ai-ticketing).
Turn on AI Ticketing
- Go to Administration → AI Ticketing.
- Choose your provider and supply what it needs so Ascent can reach it (see Choose a provider and API keys below).
- Switch Enable AI Ticketing on. This is the master switch — with it off, none of the AI features run regardless of the toggles below.
- Choose which capabilities to use:
- Auto-categorization — suggest a category and priority on each new ticket.
- Suggested replies — let technicians generate draft replies.
- Optionally pick the model to use. The field suggests a sensible default for the selected provider.
- Save.
Choose a provider
In the Provider card, pick the service that generates suggestions:
| Provider | Use it for | Base URL | API key |
|---|---|---|---|
| Anthropic (Claude) | Anthropic's Claude models | n/a | required (BYO or platform) |
| OpenAI | OpenAI's GPT models | optional (defaults to OpenAI) | required (BYO or platform) |
| OpenRouter | Many models through one account | optional (defaults to OpenRouter) | required (BYO or platform) |
| Local / Custom (OpenAI-compatible) | A server you host — Ollama, LM Studio, or any OpenAI-compatible API | required (e.g. http://localhost:11434/v1) | optional (most local servers need none) |
- The Model field's placeholder updates to a typical model for the chosen provider; type the exact model name your provider expects.
- A Base URL field appears for every provider except Anthropic. For OpenAI and OpenRouter you can leave it blank to use their standard endpoint; for a Local / Custom provider it is required — point it at your OpenAI-compatible server.
API keys
Most providers need an API key; local/custom endpoints usually do not. There are two ways to supply a key:
- Bring your own key (recommended). Paste your organization's key for the selected provider on the settings page. It is stored encrypted and is never shown again — only the last four characters are displayed as a reminder of which key is in use. Usage is billed to your own provider account. To remove it, clear the field and save.
- Platform-provided key. If your Ascent operator has configured a shared key for your provider, organizations that have not added their own can use it. When this is available, the settings page indicates so.
If you have entered your own key, it always takes precedence over the platform key. A Local / Custom endpoint can run with no key at all. If a provider that needs a key has neither a personal nor a platform key, the AI features stay inactive even when enabled.
How auto-categorization works
When auto-categorization is on, each newly created ticket is analyzed in the background. Ascent suggests:
- a category chosen from your existing ticket categories (the AI only picks from categories you have already defined — it never invents new ones),
- a priority,
- a confidence score, and
- a short reason.
The suggestion appears on the ticket's Details panel. A technician can:
- Accept — applies the suggested category and/or priority to the ticket.
- Dismiss — clears the suggestion without changing the ticket.
Categorization runs in the background and never blocks or delays ticket creation. If the AI is unavailable, the ticket is created exactly as normal, just without a suggestion.
How suggested replies work
When suggested replies are on, technicians see a suggest a reply control in the ticket reply box. Generating a suggestion drafts the next customer-facing reply based on the ticket and its conversation, then places it in the reply box. The technician can edit the draft freely before sending — nothing is sent to the customer automatically.
Only public replies are used as context when drafting a suggestion. Internal notes are never sent to any provider.
Privacy summary
- AI Ticketing is off by default and license-gated.
- No ticket content is sent to any provider unless the feature is enabled and a usable provider is configured.
- You control the provider: a third party (Anthropic, OpenAI, OpenRouter) or a server you host yourself for a fully local setup.
- Bring-your-own API keys are stored encrypted; only the last four characters are ever displayed.
- Suggested replies are drafted from public ticket content only — internal notes are excluded.
- Suggestions are advisory: a technician must accept a categorization, and a drafted reply is never sent without review.
For your operator's full data-handling commitments, see the Trust & Security section.