AI Knowledge Assistants / AI · Ready-made solution · Built in-house
Alex is the assistant in the corner of this site. Visitors ask about services, projects and prices in their own words; Alex answers from an approved set of KMCP knowledge, offers real consultation slots from our calendar and hands over to a person on WhatsApp when that is the better route.
The problem
Most questions arrive when nobody is at a desk: what do you build, have you worked in my sector, what would it cost, can we talk this week. A form collects the question and the visitor leaves. A generic chatbot answers confidently about things it does not know.
What we built
Alex is a retrieval-grounded assistant. For each reply it is given the current set of KMCP knowledge, assembled at that moment from our services, partners, completed projects and newsroom, and it answers from that. When the visitor wants to meet, it reads live availability from our calendar and books the slot. When they would rather talk to a person, the same window opens the WhatsApp line.
Built to last
Alex reads the newsroom and the project list every time, so a new article or a completed project is reflected in the next answer without a deployment. Input is capped per message, each session is rate-limited, and if the model is unavailable the visitor sees a clear message instead of a silent failure.
Easy to track
Each exchange is stored with its session and the model that answered; errors are logged separately with their cause. Bookings land in the calendar with the invite sent. The admin dashboard shows which AI readers visit the site and which Ask-AI buttons visitors use.
Built by KMCP / Inside our platform
View full size
Before every reply, Alex is handed the current KMCP context: services, partners, completed projects and the latest newsroom articles, pulled from our content store at that moment. It answers from that material rather than from general web knowledge, so what it says about us is what we have published.
When a visitor asks to meet, Alex reads free slots from our calendar, shows them in UK time alongside the visitor’s own time zone, and books the meeting with a confirmation and calendar invite.
When a message reads like a request to meet, Alex recognises it and steers the conversation to booking instead of a vague promise to follow up. Quick replies for services, process and booking keep first questions short.
The same window offers ChatGPT, Claude and Gemini with a factual, pre-written prompt about KMCP, and a WhatsApp line to the team. Alex is a front door, not a wall between the visitor and us.
An example in practice
Alex runs on kmcpsolutions.co.uk today. The same assistant can run on your website with your approved knowledge, your calendar and your handover rules.
Similar systems
