Service
AI chatbots for business
We build chatbots and conversational tools that answer questions, qualify leads and guide customers. They're trained on your brand and can live on your site or in your product, with clear limits on what they can say and do.
What's included.
- use case and scope definition
- review of source content
- conversation design and brand voice
- chatbot build, or setup of an existing tool
- website or product integration
- CRM and lead routing connections
- handoff to a person
- testing and launch checks
- conversation tracking and reporting
- ongoing review and updates
What an AI chatbot can do for a business.
A modern AI chatbot uses a large language model to understand a question in everyday language and answer using the information you give it. Older chatbots followed fixed scripts and fell apart when someone asked something unexpected.
Businesses most often use chatbots for a few jobs: answering common customer questions, helping website visitors find the right product or page, qualifying leads and booking the next step, and giving staff quick answers from internal documents.
Each job needs its own design. A support bot needs accurate answers and a clear route to a person. A lead qualification bot needs to ask the right questions and send the results to your CRM. We work out which job comes first before we build anything.
When a chatbot makes sense.
A chatbot is usually worth building when people keep asking the same questions, when visitors arrive outside business hours or when leads wait too long for a first reply. It also needs reliable source material, because it can only be as accurate as the information behind it.
It may not be the right tool if you get very few enquiries, if most questions need a person's judgment or if your product information changes constantly with no single place to keep it up to date. In those cases, a better FAQ or a simpler form may do the job for less money.
Custom chatbot or off-the-shelf tool.
Plenty of good chatbot tools already exist, and for a simple use case one of them may be all you need. A custom build makes more sense when the chatbot has to connect with your own systems, follow specific rules, match your brand closely or handle sensitive information in a particular way.
We'll tell you if an existing tool would do the job. If it would, we can help you set it up and test it properly. If it wouldn't, we'll explain what a custom build adds and what it will take to maintain.
How we keep a chatbot accurate.
The biggest risk with an AI chatbot is a confident wrong answer. We reduce that risk in a few ways.
Approved sources
The chatbot answers from content you've approved, such as your website, product documents and policies. When that content changes, the chatbot's knowledge has to change too.
Clear limits
We define the topics it can cover and those it should decline. A chatbot on a health or financial services site, for example, needs much tighter limits than one helping people pick a coffee machine.
A path to a person
When the chatbot can't answer, or when someone asks for help, it should pass the conversation to your team with the context attached.
Testing before launch
We test it with real questions, including difficult and unusual ones, and we review the answers with your team.
Review after launch
We read samples of real conversations to find wrong answers and gaps in the content. A chatbot needs ongoing care, so we include that in the plan from the start.
Data and privacy.
A chatbot may handle personal information, so we agree on the rules before connecting anything. We decide what it can read, what it can store, how long records are kept and which outside services are involved.
We also make it clear to visitors that they're talking to AI, and we follow the consent requirements that apply to your site and market. Sensitive information shouldn't go into a tool simply because a connection is available. Legal questions stay with your qualified reviewer.
How it works.
- 01
Choose the job
We look at the questions you get, where people get stuck and what your team spends time repeating. We then agree on the first job the chatbot should do.
- 02
Prepare the content
We review the information it will rely on and fix the gaps. We also set the topics it should avoid and the point where it should hand off to a person.
- 03
Build and connect
We build the chatbot or configure the tool you've chosen, then connect it to your site, product or CRM. Access permissions are agreed before we connect anything.
- 04
Test it properly
We test common, difficult and off-topic questions and review the answers with your team. It doesn't go live until the answers meet the standard we agreed on.
- 05
Launch and improve
We watch real conversations after launch, fix wrong answers and fill content gaps. We also report on what people ask and where the chatbot helps.
How we measure a chatbot.
We measure it against the job it was built to do. For support, that can mean questions resolved without a person, how often conversations get handed off, and whether answers were accurate. For leads, it can mean conversations started, leads qualified and what happened to those leads afterward.
We also keep an eye on the failures. A high number of conversations means little if people leave frustrated or receive wrong answers.
How we price it.
We scope the build in hours and confirm the applicable engineering rate before work starts. Strategy and advisory are CAD $175/h. Hosting, AI usage and software subscriptions are identified separately, because they can continue after launch.
We agree on the deliverables and review points in advance. If new work appears, we discuss the hours before expanding the scope.
Questions before we start.
How much does an AI chatbot cost?
It depends on what the chatbot needs to do and what it has to connect with. A simple setup on an existing tool costs far less than a custom build linked to your own systems. There's also an ongoing cost for AI usage, hosting and upkeep. We'll give you a scope once we know the job.
How long does it take to build a chatbot?
That depends on the state of your content, the integrations and how long reviews take. We'll give you a timeline once we've seen those pieces.
Can you train a chatbot on our website and documents?
Yes. That's usually the starting point. We review the material first, because outdated or conflicting content leads to wrong answers.
Can a chatbot qualify leads and send them to our CRM?
Yes. It can ask the questions you choose, send the details to your CRM and route the lead to the right person. Your CRM needs to allow that connection, which we'll check early on.
What happens when the chatbot doesn't know the answer?
It should say so and offer a way to reach your team. We design that handoff and test it before launch.
Can it work in French and English?
Yes, and it should be tested in each language it will use. We review answers separately in each one, because a good translation isn't always a good answer.
Will a chatbot replace our customer service team?
We don't recommend planning it that way. A chatbot can take on repetitive questions, so your team has more time for the conversations that need a person.
Related services.
Start a conversation.
Tell us what you are trying to move. We will tell you which disciplines are worth engaging, and which are not.