AI agents and chatbots that answer, qualify and book, around the clock
An AI agent is software that uses a large language model to understand a request and then take action, such as looking up an order, booking an appointment or updating your CRM. A chatbot only talks; an agent talks and does. FIT builds both, trained on your own business information, on the channels your customers already use.
What you get
- A written conversation design. What the agent handles, what it never handles, and when it hands over to a person.
- A knowledge base built from your FAQs, price lists, policies and past conversations.
- The agent deployed on your chosen channels (WhatsApp Business Platform, website chat, Instagram/Facebook DMs, email).
- Integrations with your booking tool, CRM, spreadsheet or store so the agent can take real actions.
- Human handover to your team inside WhatsApp, email or a shared inbox, with the full conversation attached.
- A test pack of real customer questions that the agent must pass before launch, and a monthly accuracy review.
Examples
Clinic appointment requests on WhatsApp
E-commerce order questions
Lead qualification for a service business
What we build
Most businesses don’t need a “do anything” AI. They need an assistant that handles the 60–80% of conversations that are repetitive (prices, availability, order status, directions, booking) and gets out of the way for everything else. That share varies by business, and we measure yours in the audit rather than guessing.
We build assistants in three levels, and most clients start at the first or second:
- Answering assistant. Answers questions from your knowledge base on one or more channels, collects contact details, and hands over to a person.
- Action-taking agent. Everything above, plus connections to your tools: it can check availability and book, look up orders, create leads in your CRM, or send a payment link.
- Multi-step agent. Handles longer processes with several steps and decisions, such as qualifying a lead, scheduling a viewing, sending the documents and following up. These always include human approval at the points where a mistake would cost money or trust.
How we keep it safe and accurate
- Grounded answers. The agent is limited to your approved information, and we write explicit “never answer” rules.
- Human in the loop. Refunds, complaints, clinical, legal and financial questions go to a person, and you choose which other topics do.
- Privacy by design. We only collect the data the conversation needs, choose AI providers whose terms exclude training on your data where possible, and document where data is stored. Our CEO is a legal practitioner, so data-protection questions get a real answer.
- Measured, not assumed. You get a monthly report covering the questions asked, how many were answered, how many were handed over, and which answers need improving.
What you need to provide
Your FAQs or past customer conversations, price lists or service menus, booking or order system access, and someone on your team who will receive handovers. We do the rest.
Typical tools
WhatsApp Business Platform (Cloud API) · Website chat widget · Instagram & Messenger · OpenAI / Anthropic Claude / open models · Cloudflare Workers · n8n or Make · Your CRM or booking system
Typical timeline
A focused single-channel agent typically goes live in 2–4 weeks, including testing with real questions.
Frequently asked questions
What is the difference between an AI chatbot and an AI agent?
A chatbot answers questions in conversation. An AI agent can also take actions in other systems, such as checking an order status, booking an appointment or creating a CRM record. Most businesses need a chatbot that can do a few well-defined actions, which is what we build.
Can the AI agent work on WhatsApp?
Yes. We use the official WhatsApp Business Platform (Cloud API) so the agent runs on your business number with Meta's approval. Under Meta's rules the bot must serve your business's own customers and use cases, not act as a general-purpose AI assistant, and that is exactly how we design it.
What happens when the AI doesn't know the answer?
It says so and hands over to a person. We set explicit rules for topics it must never answer (for example medical, legal or refund decisions) and every handover includes the full conversation so your team doesn't ask the customer to repeat themselves.
Will customers know they are talking to AI?
Yes. We always disclose that the assistant is AI. It builds trust, and several jurisdictions (including the EU AI Act and some US state laws) expect it.
How do you stop the chatbot from making things up?
The agent answers from your approved knowledge base, not general internet knowledge. We test it against a list of real questions before launch, block topics it must not handle, and review a sample of conversations every month.
How much does an AI chatbot cost to run?
There are two parts. Our one-off build fee, and running costs. Running costs are usually the AI model usage, WhatsApp conversation fees from Meta, and any hosting, and they scale with message volume. We estimate them for your volume during the free audit so there are no surprises.
Related pages
Industries
Problems this solves
Find out what you could automate, in about 4 minutes
Answer a few questions about how your business runs. You get an instant summary of your best automation opportunities and an estimate of the hours they could free up. No call needed.