Skip to main content
Once MCP is connected, your AI assistant can read your AI Sync data and answer questions directly. No dashboard, no clicking. This page is about asking well. A vague question gets a vague answer; a precise one gets a number you can act on.
Set this up first: Connect MCP. Everything below assumes your agent can already reach your account.

​
🎯 The shape of a good question

✅ Works

“What was our connect rate last week, and how did it compare to the week before?”A window, a metric, and an implied comparison.

❌ Too vague

“How are we doing?”Doing at what, over what period, compared to what?

​
💬 Questions worth asking

  • “How many dials and connects did we do today?”
  • “Is anyone online and dialling right now?”
  • “What’s our connect rate today versus yesterday?”
  • “Our connect rate dropped this week. What changed?”
  • “How many calls were answered but never reached a rep?”
  • “What share of our dials hit voicemail or got screened?”
  • “Are any of our numbers getting flagged as spam?”
That first question is the most useful one on this page. It makes the agent check the outcome breakdown rather than guess, and the answer usually points at screening, bad numbers, or pacing — not at your reps.
  • “Rank my reps by connects for the last 7 days.”
  • “Which rep has the best connect rate with at least 100 dials?”
  • “How much talk time did each rep log this week?”
Always include a minimum dial count when asking about rates. Without one, a rep with four dials and one connect can top the list at 25%.
  • “Which power list has the best connect rate?”
  • “Which lists have we barely touched?”
  • “How many bad numbers are in our lists?”
  • “How many appointments were set this week?”
  • “What’s our disposition breakdown for the last 30 days?”
  • “How many contacts are sitting in follow-up?”

​
🧠 Give your agent the vocabulary

Your agent answers better when it knows what your numbers mean. Paste this once at the start of a session:
Most of that is enforced in the product too, so the agent and the dashboard give you the same answer. The paste-in just stops your agent inventing its own definitions in conversation.

​
🔍 Checking an answer you doubt

1

Ask for the window and scope

“What date range and which reps did that cover?” Most wrong answers are right numbers over the wrong window.
2

Ask it to reconcile

“Do those outcome numbers add up to the total dials?” They always should. If they do not, something is off.
3

Cross-check one figure

Open Dialer Performance and compare a single number over the same range.
4

Watch for false zeros

A rate of exactly 0% on a small sample usually means not measured. Ask how many dials it was based on.
If your agent gives a confident percentage for a rep or list with only a handful of dials, push back. That is the single most common way a good list or a good rep gets cut on bad evidence.

​
➡️ Where to go next

Run your dialer by chat

Go beyond questions and make changes.

How calls actually work

The definitions behind every answer.

Connect MCP

Set up the account connection.

Connect the docs

Let your agent read the manual too. No login needed.

Dialer Performance

The same numbers, on screen.