The Franchise Sales Copilot lives on the org side of Franchise Sales — mainly the AI Assistant page and the Power Dialer, both scoped to your whole franchise pipeline rather than a single shop. It's really four related capabilities sharing one feature: sharpening your pipeline's existing next-best-action and deal-risk read into a specific narrative per prospect, drafting a follow-up email or call script for one prospect, giving you a short brief the moment you're about to dial someone, and turning your own free-form (typed or dictated) post-call notes into a structured outcome/next-step/objections summary. Every one of the four is built from real data already in your CRM — a prospect's profile, activity log, and the recommendation/risk signals you already see elsewhere in the app — and none of them sends anything, logs an activity, or saves anything on its own. You always review before it goes anywhere real.
Sharpened recommendations and deal-risk narratives
The AI Assistant page already shows a next-best-action and a risk score for prospects in your pipeline, computed from engagement and pipeline signals. The Copilot layers a sharper, more specific narrative on top of those same numbers — it doesn't run a second scoring pass, it just writes out a plain-prose read of whichever signals a given prospect already has. You'll see this narrative as extra text under the hero "top action" card and under other recommendation/risk cards as you scroll. If a prospect only carries one type of signal — a recommendation but no risk flag, or the reverse — the narrative only talks about that one, so it never implies a risk or action that isn't actually flagged. It refreshes along with the page's "Refresh insights" control.
Drafting outreach
From a prospect's recommendation card or deal-risk card, "Draft outreach" opens a draft for that prospect — either a short follow-up email (subject and body) or a call script (talking points only), built from their profile, their recent activity/notes log, and the same next-best-action reason already shown on the card. You can switch between the email and call-script format and it redrafts for the new channel. It's a draft only: nothing is sent and nothing is logged to the prospect's timeline from this step. You edit it, send the email or use the script on your call yourself, then log the real touch the same way you'd log any other call or email.
Power Dialer: pre-call brief and post-call summary
Inside the Power Dialer, the Copilot shows up twice around each call. As a prospect becomes your active dial target, a short pre-call brief loads automatically — a few plain sentences combining their profile, the same next-best-action/risk signals shown on the AI Assistant page, and their recent activity, so you have something to open with and know what's worth watching before the call connects. It's read-only advisory text and never writes anything back to the prospect record.
After the call, you type or voice-dictate notes into the call-notes field (dictation runs on-device in your own browser). An "Extract details with AI" action turns those notes into a short outcome label, a next step if one was mentioned, any objections raised, and a tidied summary — built strictly from your own notes, with anything you didn't mention coming back blank instead of filled in with a guess. None of that is saved by itself; it only reaches the prospect's activity log once you click one of the actual disposition buttons below the notes field, the same buttons you'd use to close out or drop a voicemail on a call either way.
Things to know
- Requires the +CRM/Franchise Sales add-on, available on the HQ and Enterprise plans, or on its own if your org purchased Franchise Sales standalone without also buying HQ.
- Stub mode: if the workspace doesn't have ANTHROPIC_API_KEY configured, every part of the Copilot falls back to placeholder output instead of a real AI read — the sharpened narrative reuses the existing heuristic wording verbatim, the outreach draft is a generic follow-up template, the pre-call brief is assembled straight from the underlying signals with no AI rewrite, and the call-summary extraction just echoes your notes back with a generic outcome label. None of that reflects a real read of your data.
- All four pieces are designed to work only from what's genuinely in your CRM — none of them fabricates details that aren't already on record. When a prospect has little or no activity logged, the pre-call brief is upfront about that and stays brief, rather than stretching to fill space with generic material.
- The sharpened narrative is additive, not a replacement — the recommendation and deal-risk numbers you already see are computed the same way whether or not the narrative loads, so it can't disagree with what's shown elsewhere on the page.
- Call-summary extraction can only work with what you actually typed or dictated. If a detail wasn't captured in your notes, the corresponding field comes back blank rather than filled in with a guess — worth a glance before you save.
Frequently asked questions
Q: Does anything in the Franchise Sales Copilot get sent to a prospect or saved automatically?
A: No. Every piece — sharpened narratives, outreach drafts, the pre-call brief, the call-summary extraction — is either read-only display or a draft you review. Nothing emails, calls, or writes to a prospect's record until you take that action yourself.
Q: What plan do I need for this?
A: It's part of the +CRM/Franchise Sales add-on. That's included on the HQ and Enterprise plans, and also available on its own if your org purchased Franchise Sales without buying HQ.
Q: How is the "sharpened" narrative different from the recommendation or risk score I already see?
A: It's the same underlying signals, just written out as a specific sentence or two instead of the shorter label already on the card. It doesn't change the recommendation itself or the risk score — it's an extra read layered on top.
Q: Can I trust the pre-call brief and outreach draft to be accurate?
A: They're built only from real data already in your CRM — the prospect's profile, activity log, and the recommendation/risk signals shown elsewhere — and are designed specifically not to invent details. That said, it's still AI-written text, so a quick read before a call or a send is worth it.
Q: What happens if I dictate call notes and the AI misses something?
A: Extraction only picks up what's actually in your notes — if something wasn't said or dictated clearly, that field comes back blank instead of guessed. You can edit the notes or the extracted fields before you click a disposition button to save anything.
Q: What does "Stub mode" mean here?
A: It means the workspace doesn't have ANTHROPIC_API_KEY configured, so you're seeing placeholder output instead of a live AI read — a generic outreach template, a brief built straight from the raw signals with no rewrite, and a call summary that mostly echoes your notes back. It's a fallback, not a real result.