Customers & Automation
Email Enquiry Auto-Draft
Last updated 19 May 2026
Email Enquiry Auto-Draft
Every inbound email to your venue address is read by AI, which drafts a contextual reply ready for you to apply, edit, or discard from your inbox. Email always drafts — it never auto-sends — so you stay in control of what goes out.
Where to find it — Web: Sidebar → Inbox → open any email conversation → AI draft chip above the composer. Mobile: Inbox tab → same chip pattern.
TL;DR
- AI drafts a reply to every inbound email automatically — no operator action required.
- Drafts appear above the composer in the inbox with Apply, Edit, Discard controls.
- 30-minute staleness warning if the conversation moved on since the draft was generated.
- Drafts only — never auto-sent. Email enquiries tend to be more nuanced; human review is the safety net.
- PRO required. Available in every market.
How it works
When an email lands at your venue's reply address ({your-slug}@reply.openchairpro.com):
- The webhook ingests the email and stores it in the conversation.
- Auto-reply detection runs — out-of-office bots and other auto-replies are skipped (no draft generated).
- AI Booking Enquiry Auto-Responder runs first — if the email is clearly a booking question, it auto-replies with available slot times (see that doc for the auto-reply path).
- Email Enquiry Auto-Draft runs on everything else (and on non-booking emails that fall through the booking-enquiry path).
- The draft writes to the conversation as a holding field on the inbound message.
- The inbox surfaces the draft as an AI draft chip above the composer.
The draft pipeline runs in the background — it doesn't block the email from appearing in your inbox. If you open the conversation before the draft is ready, you'll see the chip appear within a few seconds.
Intent-aware drafting
The same intent classifier as SMS Auto-Respond (pricing, hours, availability, booking link, general, complaint) routes the draft generator. Each intent has tailored context:
- Pricing — your service list with current prices
- Hours — your operating hours
- Availability — real slot times for the service/staff the customer asked about
- Booking link — your storefront URL with any service pre-fill
- General — your brand voice + Business Memory facts
- Complaint — no draft is generated. The email lands in your inbox unhandled — designed for you to read and reply personally.
Slot citations
When the customer asks about availability, the draft pulls real slot times and cites them in the body. The validator hard-fails the draft if any cited time doesn't exist (fabrication) or if the draft only partially cites the available slots (so the customer doesn't have to ask "what about the other times?").
If the validator fails twice, no draft is generated and the email lands in your inbox without one. Reply manually.
The draft chip
Above the composer, every inbound email with a draft shows:
- AI draft ready label (info-coloured chip)
- Click to apply — drops the draft into the composer
- Discard link below — dismiss the draft if you'd rather write fresh
After apply, edit freely in the composer. Your edits are sent verbatim. The system tracks whether you sent the draft verbatim vs edited, which feeds telemetry on draft quality (not visible in-app).
Staleness banner
If the draft is older than 30 minutes, the chip shows: "Draft {n} minutes old — review before sending". This catches the case where the conversation moved on (new messages, the customer rebooked, etc.) since the draft was generated.
You can still apply the stale draft if it's still relevant — it's a warning, not a block.
Threading
When you send the reply, the system adds In-Reply-To and References headers from the inbound messageId so the reply threads correctly in the customer's email client (Gmail, Outlook, Apple Mail).
Subject becomes Re: {inbound subject} — avoiding the Re: Re: Re: chain that some clients produce.
What gets drafted vs not
| Drafted? | |
|---|---|
| Customer asks about a service | Yes |
| Customer asks for available times | Yes (with cited slots) |
| Customer asks "Are you open Saturday?" | Yes |
| Customer says "Thanks for yesterday!" | Yes (general) |
| Customer files a complaint | No — escalates to you |
| Out-of-office auto-reply from customer | No — skipped |
| Empty body | No — skipped |
| Webhook rate limit hit (10/hour per venue) | No — skipped |
Brand voice and Business Memory
Drafts use:
- Your venue's selected brand voice persona
- Your Business Memory facts — specialities, walk-ins, atmosphere, etc.
- Your venue's avoid list — words the AI is forbidden from using
The same configuration that powers SMS Auto-Respond, Instagram DM Concierge, and notification templates also powers email drafts. Change brand voice → email drafts adjust on the next reply.
Apply flow (acceptEmailDraft)
When you Apply and Send:
- The composer text (which may include your edits) is captured.
- The system sends the email via your venue's marketing from-address with the proper threading headers.
- The outbound conversation log records
aiSendMode: "draft_accepted"or"draft_edited"based on whether you changed the text. - The draft is cleared from the conversation (the chip disappears).
If sending fails (Resend hiccup), the draft is preserved so you can retry without regenerating.
What's preserved vs cleared
| Event | Draft preserved? |
|---|---|
| Apply and Send (success) | No — cleared after send confirms |
| Apply and Send (fails) | Yes — preserved for retry |
| Discard | No — cleared immediately |
| 30-minute stale warning | Yes — the warning doesn't auto-clear |
| New inbound email on the same conversation | New draft generated — replaces old one |
On the FREE plan
Email Enquiry Auto-Draft is PRO only. Inbound emails still land in your inbox on FREE — they just don't get an AI draft. Reply manually using the standard composer.
Cost
The underlying AI generation is included in your PRO subscription. Unlike SMS Auto-Respond (1 Spark per reply) or the AI Booking Enquiry Auto-Responder (1 Spark per reply), email drafts aren't metered per-reply on the operator's Spark wallet. The cost is absorbed in OpenChair's AI cost-of-goods.
Common mistakes
| Problem | What to check |
|---|---|
| Draft chip doesn't appear after an inbound email | The draft generation has a few seconds of latency. If it's been over a minute, the email may have hit the rate limit (10/hour per venue) or been flagged as auto-reply. Reply manually. |
| Draft is wildly off-topic | Reply manually and discard the draft. If wrong drafts are persistent, update Business Memory with the specifics that should ground future drafts. |
| Customer received a reply that doesn't thread correctly | Confirm your venue has the correct reply-from address configured. The threading headers depend on it. |
| Draft says it's stale even though I just opened the email | The 30-minute clock starts when the draft was generated, not when you opened the email. If the email arrived 35 minutes ago and you just opened it, the draft is technically 35 minutes old. |
| Draft is missing for a complaint email | Drafts aren't generated for complaints by design. Read and reply personally. |
FAQ
What does Email Enquiry Auto-Draft do?
Every inbound email to your venue address is read by AI, which generates a contextual reply draft based on the message intent (pricing, hours, availability, booking link, general). The draft appears above your composer in the inbox ready for you to apply, edit, or discard. Email always drafts — it never auto-sends.
Why doesn't email auto-send like SMS?
Email enquiries tend to be longer, more specific, and harder for AI to handle reliably without human review. Drafts strike the balance — AI does the heavy lifting on tone and structure; you stay in control of what actually goes out.
How is this different from the AI Booking Enquiry Auto-Responder?
AI Booking Enquiry Auto-Responder handles emails that are clearly booking questions and auto-replies with slot times. Email Enquiry Auto-Draft covers everything else — drafts for review. Both run on the same inbound email; booking enquiries get an auto-reply, everything else gets a draft.
Do drafts cost Sparks?
No — the underlying AI generation is included in your PRO subscription. Other AI surfaces (SMS auto-respond, booking enquiry replies) consume Sparks per reply; email drafts don't. The cost is metered as part of your overall AI cost-of-goods, not per-reply.
Will my edits to the draft improve future drafts?
Indirectly. The system tracks draft_accepted vs draft_edited to monitor draft quality over time. Specific edits don't feed back into the model per-venue, but aggregate patterns inform prompt improvements. The most effective lever for improving drafts is keeping your Business Memory sharp and current.
Can I turn email drafts off?
There's no UI toggle today — drafts are generated for every applicable email on PRO. If you don't want to use them, just don't tap Apply. The draft sits in the conversation without affecting anything else.