Customer enquiries
Capture context and prepare relevant responses.
Customer & Workforce AI
Be there for the everyday questions.
Handle enquiries, bookings and routine support with approved knowledge and connected workflows. Bring a person into the conversation when judgement, sensitivity or policy calls for it.
This tends to fit when
You will recognise
at least two of these.
The signals are usually about volume and repetition rather than difficulty.
The same fifteen questions arrive every day and each is answered from scratch.
Enquiries land in four places — a shared inbox, WhatsApp, a form and the phone — and nobody has one list of them.
Good enquiries go cold at the weekend or overnight because there is no one to answer.
Two people give different answers to the same question, and both believe they are right.
Bookings and follow-ups depend on somebody remembering, and sometimes nobody does.
Everyday questions, answered from what you have approved
Help customers and teams get to the next step.
Every business has a layer of routine conversation: where is my order, can I move the appointment, what does the warranty cover, which form do I need. It arrives by email, WhatsApp, phone and the website form, and it competes with the work your team was hired to do. This system handles that layer from information you have approved, and hands the rest to a person.
BYBO builds the part between the message arriving and the next step happening. The system reads the request, finds the relevant approved information, drafts or sends a response in your tone, takes the permitted action — a booking, a status update, a ticket — and records the outcome. Where the request is sensitive, unclear or involves a commitment, it stops and routes it.
People keep the conversations that need judgement. A complaint, a pricing exception, a customer in difficulty, an employee question about their own terms: these reach a trained person with the history attached, not a cold transfer. Your team also owns the source information the system answers from, so what it says changes when your policy changes, not before.
A worked example
One conversation,
handed over cleanly.
A booking taken end to end, then the question where a person has to step in.
The conversation
Are you open Saturday?
19:31Yes — 12 to 3, and 7 to 11. Would you like a table?
19:31Table for 4 at 8
19:33Held for you. Saturday 8pm, four people. Name for the booking?
19:33And what rate can you do for 20 people?
19:41Let me get someone who can answer that properly — one moment.
19:42Written from your approved information only. It never invents an offer to keep a conversation moving.
What the system did
Answered the opening hours question from your own listing
Held a table for four at 8pm Saturday and confirmed it
Recorded the booking against the customer’s number
Stopped. This one is a discount.
Asked for a rate on a party of twenty, it does not guess. Group pricing is not in the approved information, and a promise here is a promise your business has to keep. The thread goes to a person with everything said so far attached.
Waiting on a person
Nadia, Front of HouseFull thread attached · customer told a person is coming
Take overSend a templateThe customer is told
Not left waiting on a reply that never comes. The handover is visible in the conversation.
Out of hours
The message is captured with its context and queued, rather than lost until Monday.
What it never does
Quote a price, agree an exception or make a commitment that is not in the approved information.
Illustrative example built to show the shape of the work. Not a client record.
Where it is used
Four shapes of
the same problem.
Most businesses start with one and add the second once the queue, the logging and the connections already exist.
Bookings & follow-ups
Coordinate the next step and keep a record.
Workforce assistance
Help staff find information and complete routine tasks.
Multilingual support
Test the languages and channels your audience uses.
How the work runs
Five steps.
One of them is yours.
The workflow mirrors the shape of a real conversation, from arrival to follow-up.
- 01
Enquiry
We agree the channels the system covers and how each message reaches it. Anything outside those channels keeps working exactly as it does today.
- 02
Understand
The system captures what is being asked and the context around it: who the person is, what they bought, what happened last time. Missing context is asked for rather than assumed.
- 03
Respond
It answers from approved information only, in your tone, and says it does not know instead of inventing. Permitted actions such as a booking or a status update happen here.
- 04
Escalate
Sensitive, unclear and high-impact cases go to a trained person with the full thread attached. The escalation rules are yours and can be changed in a day.
- 05
Follow up
The outcome is recorded against the customer or the ticket, and the next step gets a date and an owner, so nothing depends on memory.
Where a person decides
Sensitive conversations and commitments go to a person, with the context attached.
The line is yours to draw, and we write it down before anything runs.
Sensitive conversations and commitments go to a person, with the context attached. That covers complaints, pricing exceptions, anything contractual, and any case where the system is not confident. You set those rules and can widen them at any time — many businesses start with the system drafting every reply for a person to send, then relax that once the evaluation results give them reason to.
What you receive
5 things, and
they are all yours.
Not a demo and a slide deck. A running system, the evidence it works, and the documentation to run it without us.
A connected enquiry workflow
One path from the message arriving on any agreed channel, through understanding and response, to a recorded outcome and next step with an owner.
An approved answer source
The policies, prices, service terms and standard replies the system may answer from, with a named owner and a refresh schedule for each.
Escalation rules and a handover view
Written rules for what must reach a person, and a view that gives them the conversation, the customer record and what was already said.
An evaluation set of real conversations
Past enquiries, including the difficult ones, used to check response quality before launch and after every change to the rules.
Logs, a dashboard and training
Every conversation is inspectable, with volumes, escalations and outcomes on one view, plus training for the team who will supervise it.
What you need to provide
5 things, and none
of them technical.
You do not need a developer. You need someone who knows how the work really happens, including the exceptions nobody wrote down.
- 01
A few hundred real past enquiries, including complaints and edge cases.
- 02
The approved answers: policies, prices, service terms and what may never be promised.
- 03
A named owner for the answer source, responsible for keeping it current.
- 04
Access to the channels and systems the workflow touches.
- 05
A baseline: current response times, volumes and how many enquiries go unanswered.
What it connects to
Built around what
you already run.
Each channel and system is connected under permissions you agree first, and the system can only take the specific actions you have listed.
- Your website forms, shared inboxes and messaging channels
- Your CRM or customer records, for history and contact details
- Your booking, scheduling or field service system
- Your helpdesk or ticketing tool, so escalations land where your team already works
- Your internal policy and knowledge sources, as the approved answer set
How it is judged
5 numbers, taken
before we build.
We take a baseline from your current process first, then review the same measures after launch rather than counting messages handled.
01
Time to first useful response
not first automatic reply
02
Resolution rate: enquiries closed without a person having to redo them
03
Escalation quality: whether the right cases reached a person
with enough context
04
Booking or follow-up completion against the agreed next step
05
Review feedback from the team supervising the conversations
What drives the cost
No price here,
and here is why.
The honest answer depends on your work and your systems. Scope and fee are agreed in writing before paid work begins, and the first conversation costs nothing.
BYBO publishes no prices. Scope and fee are agreed in writing before any paid work begins, and the first conversation is free. If you are not sure whether this is the right first system, the Blueprint is a paid diagnostic that ends in a recommendation and a 90-day roadmap.
Build cost is driven mostly by channels and languages: each one adds testing, not just configuration. Running cost is driven by conversation volume and by the channel providers themselves, whose charges sit outside our fee and are shown separately in the operating plan. The cost guide sets out how to compare that against what the work costs you today.
Once
Design and build of the system
Every month
Running it, the review time it still needs, and monitoring
Before you enquire
6 questions we
are asked every time.
Answered plainly, including the ones with an uncomfortable answer.
Discuss this systemWill customers be told they are talking to a system?
Yes. We do not build systems that pretend to be a named colleague. The opening makes it clear what the customer is dealing with and how to reach a person, which in our experience reduces frustration rather than adding it. The tone is still yours, and the route to a human is always one message away.
What stops it inventing an answer?
It answers only from the information you approve, and is built to say it does not know and pass the enquiry on. Anything involving a price exception, a date it cannot verify or a commitment is routed to a person by rule. Real conversations are used to test this before launch and after each change.
What happens if it gets something wrong with a customer?
Every conversation is logged, so you can see exactly what was said. The case goes to a named owner, the wrong answer is corrected at its source, and the conversation joins the evaluation set so the same mistake is checked for at each release. Where the risk is high, that category moves behind a person.
How do you handle personal customer data?
The system reads only the fields the workflow needs, under permissions agreed in writing, and every access is recorded. Data stays in systems you own. We work to the least access that lets the workflow run, and review it as responsibilities change. Our privacy page covers how BYBO handles what we see.
Can it work on WhatsApp, voice or in more than one language?
Each of those is a scoping decision rather than a switch. Channels need provider access and their own testing; languages need their own evaluation set, because a phrase that reads well in one may not in another. We start with the channel that carries most of your volume and add from there.
What is the first step?
A free conversation about where enquiries arrive and what happens to them now. If it looks worth building, we usually start narrow: one channel, one set of questions, with a person approving replies. You can begin at apply or read how we work first.
Read more on this
Before you commit
to anything.
How to Automate Customer Enquiries Without Losing the Human Touch
To automate customer enquiries well, let the system capture, acknowledge, qualify and route. Keep judgement, empathy and commitments with people, and design the handover so no customer has to repeat themselves.
Read the article
Design the human decision into the workflow.
Where automation should pause, what the reviewer needs and how to keep responsibility clear.
Read the article
What Happens When an AI Workflow Gets Something Wrong?
Every AI workflow will get something wrong eventually. What separates a contained mistake from a damaging one is detection, a way to pause, an honest correction and a cause that actually gets fixed.
Read the articleBring us the conversation
your team has every day.
We will tell you whether it is worth building — including when it is not.