AI agents that answer SMS customer enquiries

An AI agent can read inbound texts, answer from your business knowledge and hand complex cases to your team — all from one shared inbox alongside your other channels.

8 MIN READUPDATED 8 SECTIONS

Yes, an AI agent can handle inbound SMS customer enquiries for your business. It reads each text, interprets what the customer is asking, answers from your business knowledge, and — where a question needs a person — hands the conversation to your team with the full thread intact. Dazza supports SMS natively alongside web chat, email, WhatsApp and other messaging channels, with every conversation landing in one shared inbox. For a broader look at how to evaluate AI customer service tools across every channel, see the guide to choosing an AI chatbot for customer service.

What an AI agent on SMS actually does

There is a meaningful difference between the generations of SMS response tools. The earliest sent one-way notifications. The next required customers to text a specific keyword — "STATUS" or "HELP" — and returned a fixed reply. If a customer typed something unexpected, the system could not make sense of it and either repeated the menu or fell silent.

An AI agent works differently. It reads whatever the customer actually wrote, interprets the intent, and replies in context — the same way it would in a web chat or an email thread. A customer who texts "Hi, I was supposed to get a delivery yesterday, any idea what happened?" gets a proper answer, not a prompt to type a keyword.

In practice, an AI agent connected to a business knowledge base can handle the enquiries that arrive by text every day:

  • Order status and delivery updates (with Shopify or another system connected)
  • Return and refund queries
  • Booking confirmations and changes
  • Frequently asked questions about products, services and policies
  • Triage — qualifying what a customer needs before routing to the right team member
  • After-hours coverage, with context-complete handovers waiting for the team in the morning

When the agent cannot resolve something — because it falls outside your knowledge, requires a human judgement call, or the customer asks for a person directly — it hands over with a summary of what was asked and what was already tried, so your team picks up without asking the customer to start again.

How SMS fits into a multi-channel setup

The cleanest approach to SMS automation is to run it through the same agent that handles your other channels, drawing on the same knowledge base and landing every conversation in one inbox. The alternative — a separate SMS bot with its own rules and its own inbox — means maintaining two sets of answers, two sets of logic, and no visibility across channels if the same customer contacts you by text and then by email.

Dazza connects SMS through Twilio. You connect a Twilio phone number from the Channels section of your dashboard, and from that point inbound texts arrive in the same shared inbox as your web chat, email and WhatsApp conversations. [1] The same agent, the same knowledge, and the same handover rules apply across every channel. If you are thinking about how SMS sits alongside WhatsApp and Instagram, the page on automating customer service across multiple messaging platforms covers the full picture.

One practical difference between SMS and richer channels like WhatsApp: SMS carries no buttons, images or interactive cards. Customers must type their replies rather than tap options, and a single SMS segment is capped at 160 characters. On messaging channels without interactive buttons, Dazza presents options as a numbered list rather than tappable quick replies, so customers can respond by typing a number. [2] A well-configured AI agent adapts its writing style for the channel — short, clear sentences rather than paragraph-style responses.

The role of your knowledge base in SMS answers

An AI agent on SMS is only as good as the information it can draw on. The character constraints of the channel mean responses must be concise — which puts a premium on accuracy, because there is little room for hedging or caveats. A wrong answer delivered in two lines is harder to walk back than one in a longer exchange.

Before going live on SMS, your knowledge base should cover the questions that predictably arrive by text. For most businesses these cluster around a few areas:

  • Order and delivery status — what happens after an order is placed, typical timescales, how to check progress
  • Return and refund policies — timeframes, conditions, what the customer needs to do
  • Booking and appointment information — how to change or cancel, what notice is needed
  • Contact and escalation — what to do if the answer is not here, when a person is available

Dazza reads your website, help articles, uploaded documents and pasted Q&A pairs, and stays up to date as your content changes. [3] Your own written answers take priority over website content, so if you need to override something — a temporary policy change, a specific SMS-appropriate phrasing — you can do that from the Knowledge section without touching your website. Corrections made directly from a conversation become permanent knowledge that outranks other sources.

The same principles that make knowledge management important for email apply here. The guide to automating customer support emails goes into depth on structuring your knowledge for automated responses, and most of it transfers directly to SMS.

Setting up SMS with Dazza: what is involved

SMS on Dazza uses Twilio, which means you need a Twilio account and a Twilio phone number. The setup is a one-time process: you paste your Twilio credentials into the Channels section of your Dazza dashboard, connect the number, and Dazza configures the routing itself. From that point, inbound texts to that number are handled by your agent. [4]

SMS is included from the Growth plan at £119 a month, which covers 1,500 AI conversations across all connected channels. [5] A conversation counts once, however many messages it contains, so a back-and-forth exchange of six texts uses one conversation.

Hypothetical example: a service business going live on SMS

This is a hypothetical illustration, not a real case.

Imagine a plumbing and heating business that currently handles booking enquiries by phone and email. Customers who cannot get through by phone often send a text to the same mobile number, but those messages sit in a personal inbox and get missed. The business connects that number to Twilio, adds it to Dazza, and sets up three pieces of knowledge: a pasted FAQ covering their service areas and call-out charges, a Q&A pair explaining their emergency cover hours, and a workflow template for booking a call back.

Now when a customer texts "Do you cover EH6?" at 9pm, the agent reads the question, finds the answer in the knowledge base, and replies within seconds: "Yes, we cover EH6 and surrounding postcodes. If you'd like to book a call or get a quote, I can take your details now." The customer's details land in the team's inbox, ready for the next morning. No missed texts, no separate inbox to check, no extra tool.

When and how the agent hands over to your team

Handover on SMS works the same way as on other channels. The agent transfers the conversation when the customer asks for a person, when the agent's confidence is low, when the customer repeats a question it has already tried to answer, or when a topic is configured to always go to a person. The full SMS thread, a summary of what was asked and tried, and any customer details collected transfer to your shared inbox, where a team member can pick it up without asking the customer to start again.

Because standard SMS does not carry a typing indicator, it is worth setting a clear expectation in the handover message — something like "I'm passing this to the team; they'll be back to you within two hours" — so the customer knows what to expect rather than wondering if their message was received. You can configure this in the agent's guidance rules. The dedicated page on AI chatbot human handover and escalation covers how to set triggers and test them before going live.

SMS alongside WhatsApp: which to offer

Businesses sometimes ask whether they need both SMS and WhatsApp, or whether one is enough. They serve different situations.

Factor SMS WhatsApp
Customer setup required None — works on any mobile Needs the WhatsApp app installed
Rich messages (images, buttons) No Yes
Character limit per message 160 characters per segment No practical limit
Typical use case Confirmations, simple FAQs, triage Richer support, photos of issues, orders
Requires Twilio Yes No (Meta Cloud API)
Dazza plan required Growth and above Growth and above

For most businesses, WhatsApp offers a richer experience and is worth prioritising if your customers are already on it. SMS is the more universal fallback — it reaches every mobile, requires nothing from the customer, and is a natural fit for confirmation messages and simple inbound queries. Many businesses run both, with the same agent handling whichever channel the customer chooses. The guide to WhatsApp for online stores covers that channel in detail if you are weighing up the two.

Frequently asked questions

Does an AI agent on SMS need its own separate knowledge base?

No. When SMS is set up as a channel in Dazza, it draws on the same knowledge base as your web chat, email and other channels. You do not need to duplicate or maintain a separate set of answers for SMS. Any correction or update you make in Knowledge applies everywhere automatically.

What phone number do customers text?

You connect a Twilio phone number — either an existing business number ported to Twilio, or a new number you register through Twilio — to your Dazza account. Customers text that number. Replies from the agent come from the same number.

Can the AI agent send outbound texts to customers?

Dazza's SMS channel is designed for inbound customer enquiries: the customer texts in, the agent responds. Proactive outbound SMS campaigns — where you initiate a text to a customer — are outside what Dazza does. If a customer texts first, the agent can continue that conversation across as many exchanges as needed.

How do I make sure the agent does not send unhelpful short replies because of the character limit?

Dazza adapts its reply length to the channel, but you can shape this further in the agent's guidance rules — for example, instructing it to always offer to pass to the team if a question needs more than a short answer, or to break a longer reply into a follow-up message. Testing a range of realistic questions through the preview before going live will show you quickly if the responses feel right for the channel.

Is SMS support available on the Starter plan?

SMS, along with email, WhatsApp, Instagram and phone calls, is included from the Growth plan at £119 a month. The Starter plan covers web chat only. [6]

What if the same customer contacts us by SMS and then by email about the same issue?

Each channel creates its own conversation thread in the shared inbox, so your team can see both. However, because SMS and email are separate threads, the agent will not automatically carry context from the SMS thread into the email reply — each starts fresh. If this is a pattern for your business, it is worth configuring the agent to ask for an order number or reference early in any conversation, so your team can match threads manually if needed.

Back to How to choose an AI chatbot for customer service on your website

See Dazza answer your customers from your own content. Start for free ↗