The Call Copilot — in-call answers from your records

What it is

Mid-call, a donor says: "I'm pretty sure I already gave — under my Gmail." Instead of stalling or taking their word for it, the caller asks the Copilot — and gets back, in seconds, whether that email appears in the campaign's records, and whether the gift actually went through or was a declined attempt.

The Copilot is an AI assistant built into the call screen. Look for the purple Ask the Copilot tile in the conversation rail, just under the AI brief (keyboard shortcut A). It unlocks once the call is placed.

What it can answer

The Copilot answers only from this campaign's records — its intro says exactly that: "Ask about this donor — giving history, past calls, their team. Answers come only from campaign records." It can look up:

• Donations — including attempts that failed or were declined, so "they say they already gave" gets a truthful answer either way.
• People — find someone by name, email, or phone across your calling lists, CRM, and imported donors, and pull one person's full picture: giving, suggested ask, open pledges, team.
• Call history — when this number was last reached and how it went.
• Teams — a team's goal, total raised, and leader.
• Campaign progress — the goal and total raised so far.

Every answer cites which records it checked, shown as evidence chips (e.g. Donations · 3). Answers are short — a verdict in a sentence or three — and email addresses and phone numbers in answers are masked unless the caller supplied them.

What it will never do

The Copilot is read-only by design. It cannot charge a card, send a text, edit a profile, or change anything at all — its only abilities are lookups. It also can't reach outside your campaign: it's pinned to the call it's opened on.

And it doesn't guess. If something isn't in your records, it says so; if it can't verify a claim, it tells you to treat it as unconfirmed. When the assistant is unavailable for any reason, it fails safe — the caller sees a clear message to rely on the person panel instead, and nothing is answered from thin air.

Practical limits

• Five questions per call. The tile shows how many are left; an unanswered question (a hiccup on our side) doesn't count against you.
• Each call gets a fresh thread; every question and answer is stored on the call log, so managers can review exactly what was asked and answered.
• Suggested questions get a caller started: "Have they donated to this campaign before?", "When did we last reach them?", "What's their team's total so far?"

Tip: Use the Copilot for lookups, not judgment. "Did this email ever donate here?" is a perfect question; how hard to press for an upgrade is still the caller's call — that's what the briefing, the script, and the team's notes are for.

The Copilot rides the Call Center, which is part of the Pledgix Pro and Political plans (the Pledgix team can enable it for other campaigns on request).

Related articles

• AI call briefings — a 30-second read before every call
• Call Center sessions — how calling works
• Donor data privacy — good practices