Skip to main content
usman
usman
Conversationalist
August 31, 2026
Question

Front AI

Hi guys

We have unique use case where we have certain Mentors on our platform + their respective VAs, VAs go through their inbox and support our customers. Here is the structure:

We treat their personal inboxes (Mentors) as support inboxes, the Mentees are our customers essentially. Currently mirroring the mentor convos into shared inboxes and only capturing mentee comms using our CRM integration and custom fields + rules.

Was wondering if there’s a way that we can leverage Front AI to automatically segment normal conversations from actual tickets and tag them, actual concerns raised by the mentees.

We are currently using conversation custom dropdown fields to automatically move convos to a general/no escalation inbox (e.g Conversation Type = Ticket/Conversation)

2 replies

nikolai
nikolai
Fronteer
September 2, 2026

Hi there,

Thanks for reaching out.

I believe that yes, Front should be able to support this type of segmentation, though the best setup will likely involve combining your existing CRM/custom field logic with Front’s automation and AI classification capabilities.

For example, you could use rules and existing CRM signals to identify mentor/mentee conversations, then use AI-based tagging or classification to help distinguish between normal conversations and messages that should be treated as support tickets. From there, Front can apply tags or custom fields such as “ticket,” “non-ticket,” or specific concern categories, which can then be used for views, routing, reporting, or follow-up workflows.

The key piece is defining what should count as an “actual ticket” in your workflow. If you can share a few examples of conversations that should be treated as tickets versus ones that should not, we can help validate whether AI tagging/classification would be the right fit.

    pedro_gutierrez
    pedro_gutierrez
    Conversationalist
    September 10, 2026

    Hi ​@usman Expanding on what Nikolai mentioned, in the rule, using Branch with Autopilot, you can be very specific about letting the AI know how to identify a Ticket (the key piece) from a normal conversation. You can configure it with a yes/no or multiple-choice question — for example: "Is this message raising a specific concern, issue, or request that needs resolution?" (assuming that’s how you differentiate a ticket vs conversation), and have AI route it accordingly.

    Make sure when writing the prompt, include:

    • Context: describe the mentor/mentee dynamic so the AI understands the domain

    • Category definitions: plain-language descriptions of what makes something a "ticket" vs. a general conversation

    • Tie-breaker rules: what should happen when a mentee message is ambiguous (e.g., default to Ticket to avoid missing a real concern)

    You may also want to enable "review entire conversation," so the AI doesn't mislabel a follow-up message just because it lacks context on its own.

    Always configure a "No answer matched" fallback branch to handle cases where the AI can't confidently classify, to route those to a human review step rather than letting them fall through.

    In any case, having an example would be ideal!

    Terms & ConditionsAccessibility statement