AI agent tools

An AI Reply node can do more than write a sentence back. Give it tools and it can look the contact up, tag them, move them along your pipeline, search your own documents, hand the chat to a person, book an appointment, or take an order, all inside the same conversation. Tools are switched on per node, in the node’s Built-in Tools tab. This page covers the tool surface. For the node itself (the prompt, the model, the fallback message), see AI replies.

​

Before the tools do anything

Two things work differently from where you’d expect:

• Reading your own documents isn’t available on every provider. Claude and DeepSeek can’t search documents. The AI Knowledge selector is still offered on the tab, but a step that replies on one of those two searches nothing whatever you select. Use OpenAI or Gemini if the agent needs to read your material.
• Knowledge is off by default. A new AI Reply node arrives with AI Knowledge set to No knowledge, so it answers from its prompt alone until you choose what it may search.

The AI agent is on every plan, so there is no gate to unlock. Two things about AI do depend on your plan: how many AI replies you get each month on DMLY’s shared AI, and whether you can connect your own provider key. Connect your own and replies run on it instead, outside that monthly count. See plans. An AI step can still be skipped while a flow runs, and it says why in the Inbox rather than stopping: AI replies are turned off for the workspace, neither a provider of your own nor DMLY’s shared AI is available, this month’s allowance of AI replies is used up, or the per-conversation cap is reached. See AI replies and Common mistakes.

​

CRM contact tools

Under CRM contact tools: The assistant can only do what is ticked here. Anything else your prompt asks for, it will describe rather than do. With nothing ticked, a warning says the assistant can only talk. Everything here acts on the contact in the conversation, so it needs no ids and can’t touch anyone else.

​

Worked example

A clinic’s AI Customer Service agent answers a question about laser pricing. With Get contact info it opens by name and knows this contact is already a client. With Add tags it drops pricing-enquiry on them. With Add a note it records what they asked. When they ask something clinical, Hand over to a human pauses the bot and Auto-assign to an agent puts it in front of the next practitioner. Nobody re-reads the thread; the tags and the note are already there.

​

Booking tools

Under Booking tools: Let the assistant check availability and book, reschedule or cancel appointments and classes. These act on your real Appointments data, so your Services, staff and classes have to exist first.

​

Services that are paid for up front

On a Service set to take payment before the booking, Book an appointment holds the slot for 15 minutes and DMLY — not the agent — sends the payment itself, as a Pay button showing the amount. The agent is told the amount so it can say what’s owed, but it never writes the link or the figure: it is given neither. The booking confirms on its own when the payment lands, and the hold is released if it doesn’t. That matters if you change a price mid-conversation. The agent re-reads the price every time it checks availability, so a new price is quoted from the next set of times onward, and the button always charges what the Service says right now.

​

Worked example

A salon’s booking agent has List bookable services, Check available times, Book an appointment and Hand over to a human switched on. A contact asks for a cut and colour on Saturday. The agent reads the Services, reads the genuinely open Saturday slots, offers two, books the one they pick, and hands over when they ask about a refund. Without Check available times the same agent invents plausible times and you get a double booking.

​

Commerce tools

These let the AI sell your Products and build a real order. They only work once AI ordering is turned on; see Take orders with AI. Tick them on the Built-in Tools tab under Commerce tools. The last three also need permission in AI Ordering Settings, so ticking them alone does nothing. Under Rich responses, Show product cards, Show product carousels and Show collection cards let the AI show products as cards where the channel supports it. Presentation mode can force Always use text only for this step.

​

Options the customer taps

When the agent asks a customer to pick from a few known options — a service, a day, a time, yes or no — it shows them as options the customer taps instead of typing. There is no checkbox for this: it is on for every AI Reply node, because it only changes how a reply looks, never what the agent can do. What the customer sees depends on the channel: A tap reaches the agent exactly as if the customer had typed the option, and typing still works everywhere, so someone who ignores the buttons is never stuck. The agent offers at most ten options at once; for more, it narrows the choice first — morning or afternoon, then the time.

​

AI Knowledge

The AI Knowledge block lets the agent answer from your own material (a price list, a policy, a treatment FAQ) instead of guessing: Let the assistant answer from your uploaded knowledge (files, URLs, FAQs). A Manage link next to it opens Settings → AI Knowledge, where the material lives. You build collections of sources there (files, web pages, FAQs, pasted text), then choose on the node what this agent may search:

• Uploads are PDF, TXT and MD only, up to 20 MB. There’s no CSV, DOCX or XLSX. A web page, an FAQ or pasted text needs no file at all.
• There’s no training step. A source is searchable on the next reply once its status reads ready.
• With nothing in the workspace yet, the node warns you and points you at Settings → AI Knowledge. If you leave the selector on No knowledge, there’s no warning at all; the agent just silently never searches.
• There is no Max results control on the node: how many snippets a reply retrieves is handled for you.
• When the step answers from a source that was ingested from a web address, it appends that address to the end of the reply as a plain link: at most two links per reply, and never one the answer already contains. Sources added as a file, an FAQ or pasted text have no address, so they produce no link. Because knowledge search never shows in the Tool calls trace, that appended link is often the only visible sign the agent read one of your documents.

Use it when your answers live in a document and change often: add the price list once, and every node pointed at that collection quotes the current prices. See Knowledge base.

​

Web search

Web search: Let the model search the web to ground its answer. Available on OpenAI, Claude and Gemini; not on DeepSeek. It’s off by default, and for most SMB agents it should stay off: your prices, hours and policies are not on the open web, and a searching agent is slower. Switch it on for a genuinely outward-facing question (a travel agent checking a destination’s public entry rules, say).

​

Custom functions

The Functions tab is for when a tool you need doesn’t exist. You define an HTTP call the AI can make: a name, a description (What does this function do? (helps the AI decide when to call it)), a method, a URL, headers and a Parameters (JSON Schema) body. The AI sends JSON arguments to your URL and uses the response in its reply. This is the one developer-shaped part of the AI node. A worked case: your stock lives in a warehouse system DMLY doesn’t know about, so you expose a check_stock endpoint and the agent can tell a contact whether the item is in.

​

Test before you publish

The Playground tab appears only on plans that allow your own AI provider; on DMLY’s shared AI the node shows only Prompt, Built-in Tools and Functions. The Playground runs your configuration against a test message: Test this configuration without sending a real message. Tool calls run against demo data on localhost. You get the Reply plus a Tool calls trace showing which tools the agent reached for. Nothing is sent to a contact and nothing is written to a contact record. Use it to check the agent picks the right tool. If you ask Can you check what’s on my file? and the trace shows no getContactInfo call, your prompt isn’t telling it to look anything up. The trace lists CRM tools, booking tools, custom functions, and offer_choices when the agent showed the customer options to tap. Knowledge search and web search run inside the provider, so they never appear in it; an agent answering correctly from your collections still shows an empty Tool calls trace.

​

What’s switched on by default

A new AI Reply node arrives with five tools already ticked: Get contact info, Add a note, Add tags, Hand over to a human and Move to unassigned queue (a sensible support agent). It also arrives with a full Customer Service prompt written for you, addressed from {{business_name}} (your workspace name) to {{first_name}}. The six AI templates in the template library each preload a different set: None of the six templates preselects any knowledge: they all arrive with AI Knowledge on No knowledge, so point them at a collection yourself. A template card can show Setup required (it’s still installable, you just haven’t connected what it needs yet) or Unavailable, which means the channel can’t support it. Chips like Requires AI and Requires calendar tell you which.

​

Limits worth knowing

The agent stops using tools part-way through a reply

Each AI reply makes at most eight tool-calling rounds. Past that the model is asked once more with tool calling switched off, so it always produces a reply. Separately, a mid-flow AI step (one whose exit leads to a real downstream node) that is reached more than three times in one run is skipped with a note in the Inbox: AI step skipped — per-conversation AI limit reached. A conversational AI stops replying after 40 turns and the flow moves on.

The AI isn't using a tool you switched on

Tell it to, in the prompt. Ticking a box makes a tool available; the prompt is what makes the agent reach for it. Where your plan has the Playground, check its Tool calls trace to see what it actually did.

Nothing happens and the flow just continues

The AI node never stalls a flow. If the model errors or comes back empty, it sends your Fallback message (on error / empty reply). If the step is skipped instead (no provider connected, AI replies turned off for the workspace, this month’s allowance of AI replies used up, or the per-conversation cap reached) it only leaves a note. Either way it continues out its one exit. Check the Inbox conversation for a note starting with a sparkle, and see Logs.