Website enquiry support
Answer questions about services, explain the next step and direct visitors to the appropriate form without implying that a booking or application has been accepted.
AI agents & knowledge
AI chatbots
Answer common questions, guide customers and give your team time for more complex work.
Built around a specific business need.
When the same questions arrive through your website every day, customers wait and experienced colleagues repeat information that already exists. A well-scoped chatbot gives people a useful first response, helps them find the right service and recognises when an adviser should take over. The starting point is the questions people actually ask, rather than a demonstration of everything a model can say.
Kent Consultancy would develop a bounded question-and-answer experience around approved content and clear handover rules. We would agree what the assistant may explain, which details it may collect and what it must decline. Source quality, accessibility and ongoing ownership matter as much as the underlying model, particularly when opening hours, eligibility criteria or service information change.
Who this is for
- Service teams handling repeated factual enquiries alongside more complex cases.
- Organisations with useful website content that visitors struggle to navigate.
- Internal support teams answering routine policy and process questions.
Where this can make a practical difference.
Staff information assistant
Help authorised employees locate policy guidance, with restricted material excluded from conversations where the user lacks access.
Assisted handover
Collect an agreed minimum of context and pass a concise conversation summary to a person when an answer is uncertain or the issue needs judgement.
How we approach the work.
Define the question set
Review enquiry themes and select a manageable scope. Record sensitive topics, excluded advice and escalation conditions before designing the conversation.
Prepare approved answers
Identify content owners, resolve conflicting guidance and define how updated material reaches the assistant. Where retrieval is needed, preserve links to supporting sources.
Build the interaction
Create concise responses, useful clarification questions and visible routes to human help. Keep account actions separate from general information unless identity checks are established.
Test difficult conversations
Evaluate ambiguous questions, misleading instructions, missing information and requests for personal data. Review both answer accuracy and the quality of a refusal or handover.
Launch with oversight
Introduce the assistant within an agreed channel, monitor a proportion of conversations and establish a review cycle for unanswered questions and content changes.
The sequence is tailored to the agreed scope, dependencies and risk. It describes our delivery approach, not a fixed timeline or performance guarantee.
What the engagement can deliver.
- An agreed scope and prohibited-topic register.
- Configured chatbot and approved knowledge sources.
- Conversation examples and escalation rules.
- Evaluation results and a content maintenance guide.
The proposal identifies the deliverables, acceptance evidence, responsibilities and support arrangements included in your engagement.
Measure what matters to this workflow.
Grounded answer rate
The proportion of reviewed answers supported by approved information.
Useful resolution rate
Enquiries completed satisfactorily within the agreed chatbot scope.
Handover quality
Whether transferred cases contain the context an adviser needs.
These are measures to discuss and baseline. Targets are agreed for the specific system and data; no universal improvement percentage is assumed.
What to prepare for discovery.
- A collection of common questions and current approved answers.
- Named owners for content approval and operational support.
- Clear expectations for privacy notices, retention and human availability.
Questions worth asking.
Will the chatbot answer every question?
No. Its scope should be explicit. A reliable fallback is preferable to an unsupported answer, and some questions should always go to a person.
Can it use information behind a login?
Potentially, provided the integration enforces the individual user’s access. A shared knowledge index must not become a route around existing permissions.
How is this different from conversational AI?
A chatbot is a specific assistant experience, often for a bounded question set. Conversational AI can also involve the wider design of dialogue, identity and continuity across several channels.