How AI qualifies your inquiries

Learn how Joinways AI reads your emails, Instagram and WhatsApp messages to automatically extract the key details of an event inquiry.

Every inquiry that lands in your Inquiries is analyzed by Joinways AI before you even open it. Instead of re-reading each email, Instagram message or WhatsApp by hand, you get a pre-structured card with the essentials: date, attendee count (Pax), budget and event type. The goal isn't to decide for you, but to save you the five minutes of reading and re-typing so you arrive on an inquiry that's ready to handle.

In events, speed of reply is often what wins or loses a contract. By sparing you the manual data entry, AI lets you respond faster and with reliable information, without missing anything along the way.

What you'll learn

How AI detects and reads an incoming inquiry, whatever the channel
Which details are extracted automatically and how they're pre-filled
How to read the AI summary, verify the data and keep the final say
What happens with an ambiguous message and how AI handles doubt
How your data is protected

Prerequisites

Before AI can work for you, at least one inquiry source needs to be connected to Joinways. Once the connection is made, there's nothing else to configure.

At least one connected source — Gmail, Outlook, a forwarding address, Instagram or WhatsApp (see Integrations).
No extra setup: AI analysis is on by default for every inquiry, with no option to toggle.

How AI reads an inquiry

As soon as a message reaches one of your channels, AI reads it the way an experienced salesperson would: it spots the date mentioned, the attendee count (Pax), the budget, the event type and the source channel, then pre-fills these fields on the inquiry. You only have to verify what was understood instead of typing everything yourself.

Concretely, Joinways runs four steps in a few seconds:

Detection — AI first recognizes whether this really is an event inquiry, and not a newsletter, an invoice or spam. A purely promotional email is set aside so it doesn't clutter your Inquiries.
Extraction — it then reads the content of the message and pulls out the key fields (date, headcount, budget…), even when they're buried in a long paragraph.
Inquiry creation — a card appears in your Inquiries, along with an AI summary that condenses the message into a few lines.
Confidence score — a reliability score is assigned based on the quantity and clarity of the information found, to help you prioritize.

At the end of this flow, you see an inquiry card you can read at a glance: the summary at the top, the extracted fields beside it, and the original message still accessible if you want to check a detail.

What AI extracts

AI systematically looks for the information that structures an event. When a detail is present in the message, it places it in the matching field; when it's missing, the field stays empty so you immediately spot what you'll need to ask for.

Client name and contact details (email, phone)
Event type (wedding, seminar, birthday…)
Requested date(s) and times
Attendees (Pax)
Indicative budget
Special needs (catering, accommodation, technical…)

These fields aren't just labels: the date and headcount feed your availability, the budget and event type power your reporting, and the contact details prepare your reply. A field filled in well today means accurate reporting tomorrow.

The confidence score

The score reflects how much clear information was detected. A very detailed inquiry gets a high score; a vague message ("Are you free in June?") gets a lower one because it's missing what's needed to qualify the event.

Think of it as a sorting tool, not a verdict: it tells you where to focus your energy. Handle the most qualified inquiries first, the ones ready to become events, then follow up on the others to collect what's missing.

💡 Tip: a low score doesn't mean a bad inquiry. It's often a promising lead that's simply missing a date or a budget. A well-crafted follow-up is enough to requalify it — and the score rises as soon as the information arrives.

Verify and correct the information

AI proposes, you decide. The human role stays central: you validate or correct what was understood before turning the inquiry into an event. Your corrections always take precedence over the AI's guess and are kept.

Open the inquiry from the Inquiries section: the card appears with the summary and extracted fields.
Review the AI summary and extracted fields in the side panel, with the original message in view.
Correct or complete any field if needed — for example pin down a date given as a range, or add a budget mentioned over the phone. Your edits are saved instantly.
Click Validate to convert the inquiry into an event: all verified data is carried over automatically, with no re-entry.

Edge cases

Not every message is as clean as "September 12, 120 people." Here's how AI handles the tricky ones.

Ambiguous date — for "the first weekend of June" or "this summer," AI makes a reasonable guess you can fix in one click; when the doubt is strong, it leaves the field empty rather than inventing a date.
Several dates proposed — when the client offers two options, AI flags them in the summary so you settle it together during the follow-up.
Budget as a range — a budget of "between €10,000 and €15,000" is captured as an indicative value; you refine it when quoting.
Multilingual or very short message — a message written in several languages or reduced to one sentence can lower accuracy: the inquiry is still created, and it's up to you to complete it.

Best practices

Always check the date and headcount before creating the event: these feed your availability and reporting.
Connect all your channels so nothing escapes the unified Inquiries — an unconnected Instagram inquiry is a potentially lost event.
Respond quickly to high-score inquiries: they're ready, and speed of reply is often decisive in events.
Follow up on low-score inquiries with a specific question about what's missing rather than a generic message.

Troubleshooting

An inquiry wasn't detected?

Likely cause: the channel isn't connected, or the message had no event intent. Check that the relevant channel is connected and active in Settings > Integrations. Purely promotional emails or messages with no event intent may be intentionally skipped by the AI.

The extracted information is wrong?

Likely cause: an ambiguous or multilingual message. Correct the fields directly in the inquiry panel before validating; your edit is kept and takes precedence over the AI's guess.

The score stays low despite a good exchange?

Likely cause: the key information came in by phone or in person, not in the written message. Enter the missing date, headcount and budget by hand: the score updates and the inquiry becomes ready to validate.

Real-world example

Camille runs a wine estate that hosts weddings. A client writes: "Hi, we're looking for a venue for our wedding on September 12, around 120 people, budget about €15,000." Within seconds, Joinways creates an inquiry tagged "Wedding", dated September 12, with 120 guests and a €15,000 budget — all with a high score.

Camille just checks the Barn's availability, adjusts the budget if needed, then validates the inquiry to generate the event. The time she would have spent re-typing this information she invests in a personalized reply — the one that makes the difference.

FAQ

Does the AI handle multiple languages?

Yes. The AI understands inquiries written in French, English and Spanish, among others, and extracts the key fields whatever the language of the message.

Is my data used to train a public model?

No. Your inquiries are used only to power your own Joinways workspace; they never feed a model shared with other customers.

Can I create an inquiry manually?

Yes, via the public inquiry form or by creating an event directly if the request comes in by phone.

What happens if I edit a field the AI filled in?

Your correction is saved immediately and takes precedence over the AI's guess. It's the value you validate that carries over into the event.

Can the AI get the event type wrong?

It can on an ambiguous message. The proposed type is only a suggestion: you correct it in one click in the panel before validating.