CHATBOTS

Escalate Unanswered HR Questions to a Live Agent in Teams

When the HR policy bot can't confidently answer a question from Confluence, it opens a tracked ticket in Postgres and routes the question to the HR team in Teams with full context.

CategoryChatbots
Enginesim
Difficultyintermediate
Triggerevent
Steps5
Setup~15 min

How it runs

The automated pipeline, trigger to output.

  • TriggerBot returns a low-confidence or no-match answer
  • LogicConfirm question is unanswered and not a duplicate
  • ActionLog the question and searched pages to Postgres gaps tablePostgreSQLPostgres
  • ActionPost handoff with full context to HR Teams channelMicrosoft Teams
  • OutputNotify employee a human will follow up, with ticket refMicrosoft Teams

What it does

Closes the loop on questions your Confluence-grounded bot can't handle. Low-confidence or out-of-scope questions are logged as gaps and handed to a human, giving HR a backlog of exactly what employees are asking that the handbook doesn't cover.

When to use it

When you've deployed a policy chatbot and want to capture its blind spots — questions that signal missing or unclear Confluence pages — instead of just replying "I don't know" and losing the signal.

How it works

  1. 1The bot's answer attempt triggers this flow when its confidence score is below threshold.
  2. 2A logic step confirms the question is genuinely unanswered and not a duplicate already open.
  3. 3The question, employee, and the Confluence pages that were searched are written to a Postgres gaps table with status open.
  4. 4A formatted handoff is posted to the HR team's Teams channel with a claim button context.
  5. 5The employee gets a Teams reply letting them know a human is now on it, with the ticket reference.

Set it up

What you configure once, before turning it on.

  1. 1
    Connect Microsoft TeamsChannels, chats, files.
  2. 2
    Connect PostgresAny Postgres URL — query, write, migrate.
  3. 3
    Connect ConfluenceSpaces, pages, blueprints.
  4. 4
    Set each agent's modelWe leave models unset so you pick the tier — fast + cheap, or top-quality.
  5. 5
    Tune it to your dataEdit the prompts, filters, and field mappings so it matches how your team works.
  6. 6
    Test, then turn it onRun once against a sample, confirm the output, then enable the trigger.

Run this workflow in your colony.

14-day trial. No DevOps. No Sales call. Provisioned in under a minute.