Everything explained in plain English

Section 1

What AI-Assist actually does

AI-Assist replaces the part of a receptionist that answers messages, qualifies enquiries, and books appointments — for £49 to £299 a month instead of £25,000 a year.

In practice that means: a chatbot lives on your website. When someone visits, it asks if they need anything. If they do, it has a conversation with them — answers their questions, captures their name and email, and (if they want to book) checks your calendar and makes the booking. While they sleep, while you're on a job, while you're busy with another customer.

Around that core there's a CRM (a contact list with notes), an email sequence tool (auto-emails over time), automations (rules like "when X happens, do Y"), and reports so you can see what's actually working.

Section 2

Your first week — order of operations

You don't have to set everything up at once. Here's the order we recommend, with rough time estimates:

1. 1

   Day 1 — Tell the chatbot about your business (~30 min)

   Add your hours, services, prices, and a few common Q&As. This is the "knowledge base". The chatbot uses this to answer questions accurately.

2. 2

   Day 1 — Embed the chatbot on your website (~10 min)

   Copy one line of code from the dashboard, paste it into your website. If you don't know how to do that, your web designer or our team can do it in 5 minutes.

3. 3

   Day 2 — Connect your calendar (~10 min)

   Link Google Calendar, Cal.com, Booksy, or whatever you use. Now when the chatbot books someone, it appears in your calendar like any other appointment.

4. 4

   Day 3 — Test it yourself (~15 min)

   Open your website on your phone in private/incognito mode. Have a real conversation with the chatbot like a customer would. Try to break it. Note anything weird and tweak the knowledge base.

5. 5

   Days 4-7 — Watch the leads come in

   Don't set up email sequences or automations yet. Just watch. See what real customers ask. After a week you'll know what to automate.

Once you've done the first week, look at email sequences (Section 6) and automations (Section 7). They're where the real time savings come from — but only if you set them up based on what your actual customers do, not what you imagine they'll do.

Section 3

The chatbot — making it actually useful

The chatbot is a small chat window that appears on your website. When a visitor types something, it replies. It uses Anthropic's Claude AI under the hood — the same engine some of the world's biggest companies use.

What makes a good chatbot

A good chatbot is one that knows your business. The AI itself is smart — but it can only answer questions about your prices, services, and policies if you tell it those things. That's what the knowledge base is for (Section 11).

What it can do well

• Answer common questions about your business (hours, prices, services, location)
• Capture name + email when someone wants to book or get more info
• Look at your calendar and book appointments directly
• Hand the conversation to you (via email) when something needs human judgement
• Work in plain English (and other languages — coming soon)

What it doesn't do

• Make promises or guarantees you haven't taught it to make
• Process payments directly (it points customers to your booking/checkout flow)
• Negotiate or improvise pricing
• Replace you for genuinely complex enquiries — it knows when to hand over

Section 4

Leads — what they are and how to manage them

A "lead" is a potential customer who's shown interest but hasn't bought yet. Every time someone has a conversation with your chatbot and gives their name + email, that becomes a lead in your dashboard.

The pipeline stages

Leads move through stages. Think of it like a funnel:

• New — Just came in. You haven't done anything with them yet.
• Contacted — You've replied to them or had a follow-up conversation.
• Qualified — They're a real prospect. They have budget, need, and intent.
• Converted — They became a paying customer. Celebrate.
• Lost — They went elsewhere or ghosted. Note why so you can spot patterns.

What the lead score number means

Each lead has a score from 0 to 100. Higher = more likely to buy. The AI looks at the conversation and judges things like: did they ask specific questions, did they mention a timeframe, did they say their budget, did they ask to be contacted. A score of 80+ means "follow up with this one today." Below 30 means "they were probably just curious."

Section 5

Bookings — how customers book themselves in

When you connect your calendar, the chatbot can offer real available times to customers and lock them in instantly. No back-and-forth emails to find a slot. No double bookings.

How it works

You connect Google Calendar (or Cal.com, Booksy, etc.). You tell us the kinds of appointments you offer (haircut, consultation, viewing, etc.) and how long each takes. The chatbot looks at your calendar in real time. If a customer wants a Tuesday 4pm slot and that's free, it books it. If not, it offers the nearest alternatives.

Reminders

We send automatic reminders 24 hours before, then again 1 hour before, by email and (when available) SMS. You can turn these off or change the timing. Reminders cut no-shows by a noticeable amount.

Cancellations and reschedules

The reminder emails include a link the customer can click to cancel or reschedule themselves. You don't have to play phone tag.

Section 6

Email sequences — drip campaigns explained

An email sequence is a series of emails that goes out automatically over time. Like a slow-drip irrigation system for your follow-ups. You write each email once, and from then on, anyone you put into the sequence gets all of them on schedule.

When to use one

• Welcome new leads — Day 0: thanks. Day 2: more about us. Day 5: case study. Day 10: special offer.
• Re-engage cold leads — "We haven't heard from you in 30 days — anything we can help with?"
• After a booking — Day after: how was it? Day 30: time for another?

How to enrol someone

Three ways: (1) manually add a single contact, (2) use an automation to enrol leads automatically when something happens (e.g. "every new lead goes into the welcome sequence"), or (3) bulk-import a CSV.

Section 7

Automations — the "if this, then that" engine

An automation is a recipe. It has a trigger (something that happens) and an action (what to do when it happens). Like "when a new lead comes in, send me a notification" or "when an appointment is booked, add the customer to the welcome email sequence."

The triggers we support

• New lead captured
• Appointment booked
• Form submitted (via webhook)
• Payment received
• Lead moved to a specific pipeline stage (e.g. "Qualified")

The actions we support

• Send an email (to you OR the lead)
• Add lead to an email sequence
• Send a webhook to another system you use
• Set a tag or update a lead's pipeline stage
• SMS notification (when Twilio is enabled on your account)

Example recipes

"Hot lead alert": When a lead has a score above 80 → text my mobile so I can call them within an hour.

"Onboard new customer": When an appointment is booked → send a welcome email with what to expect, parking info, what to bring.

"Win them back": When a lead is marked "Lost" → enrol them in the "30-day re-engage" sequence.

Section 8

Integrations — connecting your other tools

An integration is a connection between AI-Assist and another tool you already use. So your booking calendar, your accounting software, your video call tool, etc. don't live in separate worlds.

What we connect to today

• Google Calendar — Read/write your appointment slots
• Cal.com — Same idea, an alternative scheduling tool
• Zoom — Auto-create video meeting links for online bookings
• Booksy — Sync availability for beauty/grooming/wellness businesses
• Stripe — Accept payments at booking time
• Resend — Send transactional emails through your own domain

Coming soon (you can join the early-access waitlist)

• WhatsApp Business — Receive and reply to WhatsApp messages with the same AI
• Slack — Get lead notifications inside Slack instead of just email
• Twilio SMS — Outbound SMS reminders, missed-call text-back

How to connect one

In the dashboard, go to Integrations. Click "Connect" on the one you want. You'll be sent to that service's login screen, asked to give AI-Assist permission, and redirected back. Takes about 30 seconds. We never see your password — only the permissions you explicitly grant.

Section 9

Webhooks — the technical bit, made friendly

A webhook is just a way for two pieces of software to send each other a message in real time. Don't let the word scare you.

Imagine you have a contact form on your website that someone else built (Wix, Squarespace, a custom form). When someone submits it, you want that submission to appear as a lead in AI-Assist. A webhook is how that form "tells" AI-Assist a new submission has happened.

When do you need a webhook?

Mostly only when something OUTSIDE AI-Assist needs to send data INTO it. Examples: a Wix contact form, a Typeform survey, a custom checkout, a third-party booking widget. If everything happens inside AI-Assist (chatbot, dashboard, etc.), you don't need webhooks at all.

How to set one up (step by step)

1. Go to Webhooks in the dashboard. Click "Create new webhook."
2. Define what fields you expect (e.g. name, email, phone, message). AI-Assist will then know how to interpret incoming data.
3. Copy the URL we generate. This is the unique address that other services should send data to.
4. Paste that URL into the other service's webhook settings (every form/survey/checkout tool has a place for this — it might be called "notifications", "integrations", or "webhooks").
5. Test by submitting once on the source side and confirming a new lead appears in AI-Assist.

Section 10

Analytics — what the numbers actually mean

The Analytics page shows you a few headline numbers and some trends over time. Here's what each one really tells you:

• Conversations — How many people had a chat with your chatbot. Watch the trend, not the absolute number. Going up week over week = good.
• Leads captured — How many of those conversations resulted in someone giving their name + email. The ratio of leads ÷ conversations is your "capture rate." A healthy number is 20-40%.
• Bookings made — How many leads went on to book an appointment. Compare to leads to see your "booking rate."
• Average lead score — Mean of all your lead scores this period. If this drops sharply, your traffic quality may have changed (e.g. a new ad campaign brought in tyre-kickers).
• Most-asked questions — What topics customers care about. Use this to update your knowledge base and your website copy.

Don't obsess over week-to-week wobbles. Look at month-on-month trends.

Section 11

Knowledge base — teaching your chatbot

The knowledge base is a collection of facts about your business that the chatbot draws on when answering customers. The better your knowledge base, the better the chatbot.

What to put in it

• Your opening hours (and bank holiday exceptions)
• Your services / products and their prices
• Your address and how to find you (parking, transport)
• Your booking / cancellation policies
• Common questions you get asked daily — written out as Q&A pairs
• Anything specific (allergens, dress code, what to bring, age restrictions)

How to write good entries

Write each entry like you're explaining it to a smart friend who knows nothing about your business. Short sentences. Specific numbers. No marketing fluff. The AI is great at re-phrasing — it's bad at making things up. Give it real facts.

Section 12

Troubleshooting — common problems

"The chatbot isn't showing up on my website."

Check that the embed code is in your site's `<head>` section (not in a single page). Try in a private/incognito window — sometimes browser extensions block it. If still nothing, email us — we'll inspect your site and fix it within the hour.

"The chatbot answered something incorrectly."

Open the conversation transcript (in Conversations). See what the customer asked. Add a new entry to your knowledge base that answers that question correctly. The chatbot will use it from the next conversation onwards.

"My automation didn't fire."

Check the automation is toggled ON. Check the trigger conditions match what actually happened (e.g. if the trigger is "lead score > 80" and your test lead had score 75, it won't fire). Look at the activity log — every fire (and every miss) is logged with a reason.

"A booking double-booked me."

This shouldn't happen if your calendar integration is healthy. Check the Integrations page — is Google Calendar still connected (green dot)? If "broken" (red dot), reconnect. While reconnecting, the chatbot can't see your real availability and may book a slot that's actually taken.

"An email I expected didn't arrive."

Check the recipient's spam / junk folder first. Then check the activity log — every email send is logged with a status (Sent / Bounced / Spam-filtered). If the status is anything other than "Sent", the issue is on the recipient's side.

"I got charged but I cancelled."

Email us within 14 days at info@aiassistsmes.co.uk. UK Consumer Rights Act 2015 entitles you to cancel and get a refund within the cooling-off period. We process refunds the same day.

Section 13

Glossary — plain English definitions

Tech words you'll see in AI-Assist or in our help articles, explained without jargon.

API

The way two pieces of software talk to each other. You almost never need to touch one directly — we use APIs behind the scenes to talk to Google Calendar, Stripe, etc.

CRM

Customer Relationship Management. Just a fancy name for "contact list with notes about each person." Your Leads + Clients pages together are your CRM.

Embed code

A small snippet of code you paste into your website to make the chatbot appear. Looks scary, isn't — it's one line.

GDPR

UK + EU law about handling people's personal data fairly. We're fully compliant. You don't have to do anything special — we built it in.

Lead

A potential customer who's shown interest. Has a name and a contact method but hasn't bought yet.

OAuth

The "sign in with Google / Facebook / etc." flow. When you connect Google Calendar, OAuth is what you're doing — granting AI-Assist permission without sharing your password.

Pipeline

The stages a lead goes through from first contact to either becoming a customer or moving on. Think of it as a conveyor belt.

Webhook

A way for one piece of software to instantly notify another that something happened. We use this to receive form submissions from outside AI-Assist. See Section 9.

SaaS

Software as a Service. Software you use on a subscription instead of buying. AI-Assist is SaaS; so are Microsoft 365, Netflix, Stripe.

Trigger / Action

In an automation: the trigger is the "if" (something that happens), the action is the "then" (what to do). See Section 7.