What Does an AI-Managed Enquiry Workflow Actually Look Like?

"AI for enquiries" is one of those phrases that gets used a lot and explained rarely. Most business owners who hear it picture either a clunky chat widget or something vaguely magical. Neither is accurate. So here is a plain walkthrough of what an AI enquiry workflow actually does for a service business, step by step, once it is running.

In short: A managed enquiry workflow has five steps — capture, acknowledge, qualify, human review, and handover. The output is not a logged contact; it is a qualified lead with full context, ready for your team to act on.

The example below is the shape of the workflow we run for clients. The details vary by business, but the steps do not.

Step 1 — capture: nothing waits, nothing is missed

An enquiry arrives. It might come through your website contact form at 9.40pm on a Tuesday, an email from a portal listing, a message left over the weekend, or a WhatsApp from someone who found you on Google. The first job of the workflow is simple but unforgiving: log every enquiry, from every channel, the moment it arrives.

This matters more than it sounds. In most service businesses, enquiries live in three or four separate inboxes, and the ones that arrive outside office hours sit unread until someone checks. Capture means nothing waits, nothing is missed, and every enquiry enters the same process regardless of where it came from or what time it landed.

Step 2 — acknowledge and answer: a real reply, not a receipt

Within moments of capture, the enquirer gets a real response — not a "we have received your message" auto-reply, but an answer that engages with what they actually asked.

The response draws only on an approved knowledge base: your services, your coverage area, your process, your availability, whatever you have signed off. If the enquirer asks something the knowledge base does not cover, the workflow says so honestly and routes the question onward rather than improvising. That constraint is deliberate. The workflow never guesses on your behalf.

Step 3 — qualify: turning an enquiry into a lead

A polite acknowledgement is not a lead. The next job is to ask the questions your team would ask if they had picked up the phone: what does the enquirer need, when do they need it, where are they, what is the rough scope or budget, and how do we reach them?

This happens conversationally, not as a form to fill in, and it is the step that transforms the output. An unqualified enquiry is a name and a maybe. A qualified one tells your team whether it is worth a call, how urgent it is, and what to say when they make it.

Step 4 — human review: the checkpoint before a message reaches a client

Not everything goes straight through. Certain categories always stop for a human before anything is sent or actioned: complaints, anything touching legal, financial or medical territory, messages that are ambiguous or unusual, and high-value enquiries where the stakes justify a second pair of eyes.

The reviewer sees the conversation and the drafted next step, and either approves it, edits it, or takes the enquiry over entirely. This is not a safety net bolted on at the end — it is a designed checkpoint, and it is the reason the workflow can be trusted with real customers. We have written more about how this works in practice.

Step 5 — handover: a brief, not a notification

Finally, your team gets the result: a structured summary of who enquired, what they want, when they want it, the answers collected during qualification, and the full conversation if they want the detail. It arrives where your team already works — email, CRM, or both.

The difference from a normal enquiry notification is that there is nothing left to chase. The lead arrives ready to act on, and the first human touch is a productive one: a call to someone whose need, timeline and details are already known.

What this is not — and why that matters

It is worth being precise, because the market is noisy. This is not a chatbot widget you bolt onto a website and configure yourself. It is not a script that sends the same three messages to everyone. And it is not a phone tree with better grammar.

A chatbot is a tool you buy and then babysit. A managed workflow is a service: it is designed around your business, run and monitored by people whose job is to keep it accurate, and adjusted as your services, prices and seasons change. When something is off, it is our job to fix it — not yours to notice. You can see the full set of workflows we run on our services page.

What actually changes once it is running

The day-to-day difference is undramatic, which is rather the point. Every enquiry gets a fast, accurate first response, whatever the hour. Your team stops triaging inboxes and starts each conversation with a qualified lead and full context. And the enquiries that used to quietly disappear — the Saturday-night form fill, the message that landed during a busy Monday — stop disappearing.

That is the whole picture. No magic, no hype: a defined process, run reliably, with human judgement built in where it belongs.

Frequently asked questions

How is a managed workflow different from a chatbot?

A chatbot is a tool — you buy it, configure it, and manage it yourself. A managed workflow is a service: it is designed around your specific business, monitored by people whose job is keeping it accurate, and updated when your services or pricing change. The output is also different: a chatbot typically logs a contact; a managed workflow delivers a qualified lead with full context.

Which channels does the workflow cover?

The workflow handles enquiries from website contact forms, email, WhatsApp, and direct messages. Coverage depends on which channels your business uses — this is set up during onboarding based on where your enquiries actually arrive.

Can I review what the workflow has said to my customers?

Yes — everything the workflow sends is visible in a shared log, and you receive a full conversation summary with every qualified lead. There is no black box. If the tone or content ever needs adjusting, that feedback goes directly into the knowledge base.

What happens if the workflow receives an unusual or sensitive enquiry?

Any message that triggers a defined review category — complaint, regulated topic, high-value, ambiguous — is stopped before sending and placed in a review queue. A person reviews the conversation and the drafted response, then approves, edits, escalates, or takes over. The workflow never acts on sensitive items without sign-off.