Reply with AI

The AI Reply step lets an assistant read the conversation and write the reply, instead of you scripting every message. You give it instructions, pick a provider and model, and choose what it is allowed to do. It works on every channel your automations run on. Two things to know before you start:

• It works before you connect anything. A new AI Reply step answers on DMLY’s own shared AI right away; you don’t need a provider account to try it. Connect your own OpenAI, Claude, Gemini or DeepSeek key under Integrations for unlimited, un-metered replies on your own account and full control over the model, if your plan allows it.
• Replies on the shared AI count against your plan. Every plan includes a set number of AI replies a month on DMLY’s built-in AI; replies on your own connected key are never counted. See AI Replies and your own AI keys below.

​

Add the step

​

The prompt

The Prompt tab is where you tell the assistant who it is and what it should do. System prompt / instructions carries the whole personality. The prompt a new step starts with is a working customer-service assistant covering its role, greeting, what it does, its identity, staying grounded in what it knows, its tools, its style, and what to do when the question is off-topic. Edit the parts that are wrong for your business (usually the greeting, what you sell, and the off-topic rule) and leave the rest. You can use tokens in the prompt. {{business_name}} becomes your workspace name, and {{first_name}} becomes the contact’s first name, so the same prompt reads correctly for everyone. The rest of the tab handles the edges of the turn:

• Fallback message (on error / empty reply): what gets sent if the provider errors or returns nothing. The default is Sorry, I couldn’t process that just now — a team member will follow up shortly.
• Follow-up messages: extra messages sent after the AI reply, every time. Good for a fixed closing line; bad for anything the AI might have already said.
• Typing delay before reply (seconds): a pause before the reply lands, so it doesn’t arrive instantly. It is capped at 5 seconds however large a number you type, and your workspace can set a floor for it under Conversation settings.

​

Provider and model

The Configuration tab decides which service answers, but it only appears when your plan allows connecting your own AI provider key. If it doesn’t, the tab is hidden and the step runs on DMLY’s shared AI instead, whose model is decided for you; skip ahead to AI billing below. Provider lists only the AI integrations you have actually connected. Model is a free-text box with suggestions: pick one from the list, or type any model id your provider supports. All four can call tools and custom functions. Those are the only four providers the AI Reply step can use. Reasoning effort only appears when you pick OpenAI; no other provider reads it. DeepSeek offers a deepseek-reasoner model, but the effort setting still doesn’t apply to it. Two more settings:

• Max tokens: roughly how long a reply can get. New steps start at 500.
• Conversation history (messages): how much of the chat the assistant sees. New steps start at 12, and the history is capped at 50 messages however high you set it. Only real messages count; internal notes are never shown to the AI.

​

Tools and knowledge

The Built-in Tools tab is where you decide what the assistant can do (look up a contact, add tags, hand over to a human, check availability and book an appointment), and the Functions tab lets it call your own URLs. Both are covered in AI tools and functions. Two settings on that tab are worth knowing about here:

• AI Knowledge lets the assistant answer from your own material. New steps arrive on No knowledge and do nothing until you choose All workspace knowledge, Selected collections or Selected sources. You build the collections under Settings → AI Knowledge, which works on DMLY’s shared AI as well as on your own key. A step replying on Claude or DeepSeek searches nothing whatever you select, because neither can read documents. 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. Full detail in Knowledge base.
• Web search lets the model search the web to ground its answer. Not available on DeepSeek.

​

Test before you publish

The Playground tab runs your configuration against a test message and shows the Reply plus any Tool calls it made. Nothing is sent to a contact and nothing is saved to the Inbox; tool calls run against demo data, so a test that “adds a tag” doesn’t touch a real contact.

​

When it keeps chatting, and when it doesn’t

The step works out for itself whether the conversation should continue:

• If the AI Reply step leads to an End step, a note, or nothing at all, it stays in the conversation and answers the contact’s next message too. This is what you want for a support assistant.
• If it leads to another step, it replies once and the flow moves on.

If the assistant hands the conversation to a human during its turn, it stops immediately rather than talking over your team.

​

AI Replies and your own AI keys

Every plan can add, publish and run the AI Reply step; there’s no tier where AI is switched off. What differs by plan is where the reply comes from, and whether it’s metered:

• DMLY’s shared AI answers by default; it’s what a fresh AI Reply step uses until you connect your own provider. Replies on it count against your plan’s AI Replies allowance for the month.
• Your own AI key (OpenAI, Claude, Gemini or DeepSeek, connected under Integrations) replies unlimited and un-metered, but only if your plan allows connecting one. If it doesn’t, the provider tiles stay visible with a Plan locked badge, and opening one shows This plan uses the built-in system AI. Upgrade to connect your own AI provider keys.

Your plan’s AI Replies allowance, and whether it allows your own keys, are shown alongside your other allowances when you pick or change plan. See Plans. There’s no running “used this month” counter elsewhere in the app; the first sign you’ve reached the cap is the conversation note described below.

What the step looks like on the shared AI

When your plan doesn’t allow your own keys, the step’s Configuration and Playground tabs disappear; only Prompt, Built-in Tools and Functions remain, and the model and parameters are managed for you. AI Knowledge, Web search and the CRM and booking tools all still work: knowledge on the shared AI is backed by the same built-in account that writes the reply, so a collection you build under Settings → AI Knowledge is searched normally.If your plan does allow your own keys but you simply haven’t connected a provider yet, the tabs stay visible; you’ll see an orange No AI provider is connected banner at the top of the step instead, and it keeps running on the shared AI until you connect one.

Reaching the monthly AI Replies limit

Once the shared AI has answered as many times this month as your plan allows, the next AI Reply turn is skipped instead of failing. A system note reads AI reply limit reached for this month — it resets at the start of next month., followed by Connect your own AI provider keys for unlimited replies, or upgrade the plan. If your plan doesn’t allow your own keys, that follow-up is just Upgrade the plan for more AI replies. It resets automatically at the start of the next calendar month; there’s nothing to do on your end. Replies on your own connected key are never subject to this cap.

Your own key stopped working after a plan change

If your plan moves to one that doesn’t allow your own AI keys, a key you already connected stays saved but stops being used. The AI Reply step doesn’t stop, though; it falls back to the shared AI automatically, so your contacts keep getting replies, just metered against the AI Replies allowance instead of running on your own account. Move back to a plan that allows your own keys and the same key works again; you don’t need to reconnect.

​

When the AI doesn’t reply

The flow never gets stuck on an AI step. If something is wrong, it moves on to the next step rather than failing.

A system note says the AI step failed or was skipped

A note starting AI step failed — means the provider errored or returned an empty reply. What follows is the reason: the provider’s own error message, or the provider returned no reply when the response was simply empty. If your Fallback message was set, (fallback message sent) is appended and the contact received it. See the next section below. If it was empty, the contact got nothing at all, and this note is the only sign anything happened.A few other notes look similar but mean something different: AI reply limit reached for this month is the AI Replies cap. See AI billing above. AI step skipped — no AI provider is connected is rare: DMLY’s shared AI normally answers automatically even before you connect anything, so you’d only see this if that failed too.

The contact got the fallback message instead of a real answer

The provider errored or returned an empty reply, so your Fallback message was sent. This is usually a bad model id, a provider outage, or an expired API key. Check the provider under Integrations and try the same message in the Playground. If you have cleared the Fallback message, nothing is sent to the contact at all; you get the failed note above instead.

The AI answered a couple of times, then stopped

There are two limits, and a step is subject to one or the other, never both. Which one applies depends on where the step sits, exactly as described under When it keeps chatting, and when it doesn’t above.A mid-flow step, one whose output leads to another step, can fire three times in a run. That’s a guard against a loop back into the step running up your provider bill. Past three, the step is skipped and a system note reads AI step skipped — per-conversation AI limit reached. If you see that note on a normal flow, check for a loop back into the AI step.A conversational step (one whose output leads to an End step, a note, or nothing) answers up to 40 turns in a run instead, and never gets the three-call limit. At 40 the step moves on instead of answering again, and leaves a system note: AI conversation reached its reply limit — a teammate should take it from here.

The AI ignores your documents

The AI Knowledge selector on the Built-in Tools tab is still on No knowledge, which is where every new step starts. Choose All workspace knowledge (or pick collections) and check the material is there under Settings → AI Knowledge. Two other causes: the step replies on Claude or DeepSeek, which can’t read documents at all, or the sources are still indexing. See Knowledge base.

For anything else, see Automation not triggering and Common mistakes.