Sales · Inbound module
We don’t trust a bot to talk to our customers
Afraid an AI chat will promise a price, invent a date or annoy a loyal customer? What Hue is allowed to say, what it refuses, and when a person takes over.
Starts on Free
In your words
We are a family-run plumbing and heating wholesaler in Halifax, three generations in. Our counter people know half our customers by first name. When a contractor calls, he gets someone who knows his trucks, his jobs and his habits. That is why he buys from us and not from the big box down the road.
Every few months someone pitches us an AI chatbot. We tried one years ago and it told a customer we stocked a boiler we had never carried. Our worry is simple: one wrong answer, said confidently in our name, and a relationship we built over twenty years takes a hit we can’t see coming.
Why it happens
The fear is reasonable. A general chatbot is built to keep talking, so when it does not know, it fills the gap with something plausible. It answers questions it should hand over, it does not say it is a machine, and nobody at the company reads what it said until a customer complains.
The real question is not “bot or no bot” but “what is it allowed to say, and who checks?”. Most tools leave that vague: no list of what it may answer, no rule for when to stop, no record a manager can read. Without those, the only safe choice looks like keeping it off the site entirely.
What it costs
Keeping the site silent has its own price. The contractor who checks at 6 a.m. whether you have a fitting in stock, before his crew leaves, gets no answer and drives to whoever is open. Loyal customers forgive a lot, but new ones never get the chance to become loyal.
And the questions still arrive. They land on the counter, by phone, all day, and the people who know the customers best spend their time on opening hours and delivery zones instead of on the relationships you are afraid to lose.
How Humonix fixes it
01Agent
It answers only from what you wrote
Hue reads your fact sheet, your own notes and your website pages, nothing else. If the answer is not there, it says it does not know and offers to pass the question to your team. It never invents a price, stock, a delivery date or a warranty.
02Agent
It says what it is
Hue says it is an AI assistant. On texts and social messages, that line is always there, whatever the settings; on the website chat it is on by default.
03Agent
It knows when to stop
Before answering, Hue checks the message for a request for a person, a sensitive topic, the same question asked again, or an upset tone. When it sees one, it promises nothing (no refunds, price exceptions, legal or safety advice) and says a person is checking.
04Agent
A person sees what it wasn’t sure of
Every answer gets a confidence score with its reasons. Below your threshold, the conversation goes to “Needs a human” with Hue’s draft attached, assigned to the person you name, and you can report any answer you don’t like.
05Agent
Everything is on the record
Every conversation is kept and readable by your team. When a person writes a better answer, the owner can save it so Hue uses that wording next time.
01Agent
The department
On Hue Free and Hue, this is the chat on your website. With Hue Pro, the same rules follow Hue onto texts and email, and the Inbound module adds an Inbox with response times.
02Hue
Hue’s part
Hue never asks for passwords, card numbers or government IDs, and it ignores a visitor who tries to change its rules by typing instructions into the chat.
03You
Your yes
You approve the fact sheet and the greeting before Hue goes live. Hue can be wrong; that is why it shows where its answer came from and stops for a person, and why an owner decides which kinds of answers it may give alone.
What you see
The “Needs a human” list on a sample day: two conversations Hue wasn’t sure of, each with the reason, Hue’s draft and the person assigned.
Where to start
Free
One free AI visibility check and Hue Free on your site with a small badge. No card.
Questions
Will customers know they’re talking to an AI?
Yes. Hue says so. On texts and social messages the line can’t be turned off; on the site it is on unless you decide otherwise, and even then Hue says it is an AI when asked.
What if it gives a wrong answer anyway?
It can happen; Hue can be wrong. Each answer carries its sources, low-confidence ones go to a person, and when you find a mistake you correct the fact once and report the answer so it is reviewed.
Can we try it without putting it in front of customers?
Yes. Hue Free is built from your public pages and you approve it before it answers anyone. You can ask it your own customers’ questions first and read how it answers.
Who sees the conversations?
Your team, in your client space, hosted in North America (Toronto, Canada). They are your records: you can export them, and they are deleted with your account.