Ad-hoc Question Agent

Ask a question in Slack and get a direct answer in the same thread, with the supporting numbers and the query used to find them.

Trigger: an executive asks a question in Slack
AI steps: answer the question using the requested opportunity data
Output: the answer, supporting numbers, and query posted in the original thread

See Agent Template

Copied

Ad-hoc Question Agent: an executive asks a question in Slack; answer the question using the requested opportunity data; the answer, supporting numbers, and query posted in the original thread

How it works

Agent breakdown

A blueprint to adapt, not a fixed recipe. Set the trigger, thresholds, and methodology that fit your team, and each step builds on the last, using the conversation and CRM context you already capture.

Answer where the question was asked

Trigger

Record-Triggered · An executive asks in Slack

Pull only what was asked

Action

Get/Lookup · Opportunity · Scope: whatever the question asks

Answer with the data

Action

AI Step · Output: the answer and its data

Reply in the thread

Output

Send Slack Message · Channel: the thread it was asked in

TriggerStep 1 of 6

Answer where the question was asked

When an executive asks a question in Slack, the agent picks it up in the same thread. Nobody has to turn the question into a separate request or remember to come back with the answer later.

Trigger: An executive asks a question in Slack
Location: The Slack thread where the question was asked

ActionStep 2 of 6

Pull only what was asked

Next, the agent uses the question to decide which opportunity fields to pull. It only retrieves the data needed for that question, so the executive gets a focused answer rather than a standing report or dashboard.

Fields: Taken from the question
Excludes: Standing reports
Limit: One question per thread

ActionStep 3 of 6

Answer with the data

The agent answers the question directly in under 100 words, then provides the numbers behind the answer and the query it used. If the question is ambiguous, it states which reading it took rather than answering multiple versions of the question.

Prompt

Answer the executive's question using the opportunity data pulled from the question's scope.

Start with a direct answer in under 100 words. Then provide the numbers behind the answer and the query used to retrieve them.

Use only the requested opportunity data.
Keep the answer focused on the question asked.
If the question is ambiguous, state which reading you took and answer only that reading.
If you cannot read the requested data cleanly, do not guess. State what is unclear so the question can be escalated to RevOps.

Return the direct answer, the supporting numbers, and the query used.

Copy

OutputStep 4 of 6

Reply in the thread

The agent posts the answer in the Slack thread where the executive asked the question, with the query attached so anybody can check how it was calculated. If the requested data could not be read cleanly, the agent routes the question to RevOps instead of guessing.

Includes: The query that was run
Escalates: To RevOps if unclear

Output

The answer, supporting numbers, and query in the original Slack thread

Get a direct answer where you asked the question, followed by the numbers behind it and the query used to retrieve them. Anybody in the thread can check the data, while questions the agent cannot read cleanly go to RevOps instead of being guessed.