The chat and your agent

Each client has a single conversation with you. By default your agent handles it; you see the whole thing in the app, you can step in whenever you want, and the agent notifies you when it cannot go on.

How a conversation starts

When someone opens your link and taps Chat with <your name>, they see a greeting from your agent and a form asking for their email. We send them a six-digit code; until they enter it, they cannot write. That way nobody spends your credits without giving at least one real contact, and the client can come back to their conversation from any device with that same email.

If they had already talked to you, they get their conversation back; if not, a new one starts. You see it appear in Chats right away.

What the agent does

• Answers in your tone and with what you told it about yourself, your styles and the way you work.
• Asks what it needs to understand the tattoo: idea, size, placement, style, references.
• When the client wants to book, it shows free slots from your calendar across three days, two per day, as long as one session of the project. The project's duration is the one you set; otherwise the chosen flash's; otherwise the one Drop estimates from reference hours by body zone, size and style, at your pace going by your flash durations (it needs the size and the zone or the style). Whatever goes beyond your session length is split into equal sessions, or into the planned sessions you set. Each session confirmed or done is taken off, and when nothing is left, the slots last your default session. The client can ask for others.
• When the client picks one, it holds it and tells them you will confirm. It never confirms an appointment or says it is doing so: you do that with Confirm or Reject on the card in the chat or on the booking page. If you reject it or the deadline passes, it offers other slots.
• Before wrapping up, it shows them a project summary so they can confirm or correct it.
• If the client has several tattoos in progress with you, the agent understands which one they mean and, if it is not sure, asks.
• The design is yours: the agent never draws or describes a design of its own, gives no opinion on a tattoo, a design or a reference (not even a compliment) and suggests no changes. If the client asks, it says you will look at it, writes down what they want and carries on. It is no reason to hand the conversation to you.

When you mark a project as completed, it is Drop (not the agent) that leaves the review request in the chat: see Reviews.

The agent never makes things up: it only states dates, slots and details it has just checked. If something changes in a project or a booking (because you edited it in the app, moved an appointment, or a held slot expired), a notice appears inside the conversation, in the client's language, for example “Your appointment is confirmed” or “Your project has changed”, with what changed, and the agent takes it into account in its next reply. The client sees it instantly, without reloading. The details the agent itself notes down while chatting do not produce a notice.

When it notifies you

You get a notification when:

• The agent passes the conversation to you (handover). It is the last resort: it only does so when the client wants to cancel or change something that is your call, when they insist on talking to you after the agent offered to help, when your “When would you rather answer yourself?” rule says so, or when the client cannot go on without an answer only you have.
• A message arrives that the agent is not going to answer: you have it switched off in that conversation, it has handed the conversation to you, or it is paused for possible spam. If the agent is going to answer, we do not bother you, and neither if the conversation is archived.
• A conversation is paused for possible spam (see Clients who abuse the chat).
• You run out of credits: the notification says it is because of credits.

After a handover the agent goes quiet in that conversation: its switch stays on AI off and the client sees the same as when the agent is switched off. As soon as you write in the conversation, or switch it back on, the agent picks up again and takes into account what you told the client.

The conversation is marked Needs you (“Your turn”) until you reply. If no reply is needed, swipe the card to the left and tap Mark as read: it stops counting as pending without sending anything to the client, and the agent stays as it was (if it was paused or had handed the conversation to you, it does not answer again). If the client writes again and the agent is not going to answer, it goes back to Needs you. In Chats you can filter by All, Needs you, AI off, New (those from the last seven days) and Archived, and the tab shows a number with the ones waiting for you. The Drop icon on your iPhone shows that same number, even with the app closed, if you give it permission for notifications. The search finds a conversation by the client's name, email or phone, with or without accents (“garcia” finds García).

If your credits drop below the threshold, you will see the Low credits notice above the list; tap Dismiss to hide it until you open the app again.

Archiving a conversation

To get a conversation out of the way without deleting it, swipe its card to the left and tap Archive. Archiving it switches the AI off in that conversation and leaves it silent: if the client writes again, it stays archived, you get no notification for their message, and it does not count in Needs you, in Waiting for you on Home, or in the number on the tab or the icon. Notices about bookings and held slots still reach you, because they belong to your calendar. Nothing is deleted: the whole history stays.

Archived conversations leave All and the other filters, and you find them in Archived, at the end of the filter row. The search in All finds them too, with the Archived badge. To get one back, swipe its card and tap Unarchive (the AI stays off until you switch it on; if the client wrote while it was archived and you have not replied, it goes back to Needs you), or switch its AI switch on: switching it on also unarchives it.

Switching the agent off or on in a conversation

In the chat header and on the card in the list you have the AI on / AI off switch. Switch it off to run the conversation yourself; switch it on so the agent picks it up with all the context. The change is saved a couple of seconds after the last tap, and if you change it more than three times in a minute we ask you to confirm. While it is off, the client sees “<your name> will answer as soon as they can.” and each message of theirs leaves the conversation in Needs you.

What new conversations start with is up to you in Settings → AI assistant, under AI in new chats (on by default). If you switch it off, we ask you to confirm: new chats will start with the AI off, and the ones you already have do not change. With it off, the client still sees the greeting and each message of theirs reaches you as a notification. Switching it on asks for no confirmation.

Replying yourself

Write as in any chat. When the agent hands a conversation to you, it leaves you a suggested reply for the client, in gray inside the text box and in full in the AI suggestion strip above it (if it is very long, scroll inside the strip to read the rest): tap Use to have it typed into the box so you can go over it before sending, or write your own. The send button always sends. You can send up to six photos at a time, also by pasting them or dragging them from another app. With the quick-action buttons you send the same elements the agent uses: Send slots and Send confirmation of the project. Before they go out you see how the client will receive them and you can change them: in the confirmation, any detail of the project (it is saved to the project when you send); in the slots, how long the session lasts (from 1 h to 8 h, or what is left of the project; the slots are recalculated and, if one you had chosen no longer fits, the app unchecks it and tells you). If the project has no duration, the app estimates it from its size, zone and style, proposes slots for that session and shows the Estimate below with what it comes from, so you can change it if it does not fit and which ones you propose among the free ones in the next 20 days, up to six across three days and two per day. With Choose another day you open a month calendar with a dot on the days that have a slot; when you tap one, its slots are added to the list. If you change the duration, it applies to that session: the slot the client holds lasts that long and the project keeps its total duration. If the project's planned hours are already done, the sheet tells you and offers Complete project. If you touch nothing, the usual proposal goes out, and if a slot you chose gets taken in the meantime, the app tells you and sends nothing. Sending slots holds none of them, and sending one of these elements does not switch the AI back on in a conversation the agent handed to you: that takes you writing or the switch. While a booking waits for its deposit there is also Send deposit, and while the consent is unsigned, Send consent: you see the card as the client will receive it, and sending it gives it to them again, with its email (once an hour; if it was just sent, the sheet tells you when you can).

If the conversation has several projects, when you tap one of those buttons you choose the project in a menu that opens over the button itself.

Your messages carry the You label and the agent's the AI label; the client's go on the left, with no label. Changes to projects and appointments appear as notices in the center, so you always know who said what. Proposals already answered, or with a more recent one below, appear disabled.

Saving chat photos to your iPhone

You can save any photo from a conversation to Photos, whether the client's or yours. Tap it to see it full screen and tap the save button, bottom right; or hold your finger on the photo in the chat and choose Save to Photos. When it is saved you will feel a vibration and see Saved. With Share, bottom left of the viewer, you can send it to another app or use Save Image.

The photo is saved at a good size, not the chat thumbnail. If the client has just sent it and it is still being prepared, the app tells you: try again in a moment.

The first time, the iPhone asks you for permission for Drop to add photos; Drop only adds, it never sees your photo library. If you said no, the app reminds you with Open Settings so you can change it.

What the client sees

• Your name, a permanent notice that they are talking to an AI, and a button to see your Flash without leaving the chat.
• Who is writing to them at each moment: your messages carry your name and your photo, and the agent's carry the <your name>'s AI assistant label with its own icon. Consecutive messages from the same author are grouped, with the name on the first and the time on the last.
• Quick-reply suggestions to answer with one tap.
• The slots and the project summary as buttons, with the project's title above them: they tap a slot to hold it, Ask for other slots if none works for them, and Looks right or Correct it on the summary. What they choose reaches the agent as exact data, not as text to interpret. A proposal already answered, or with a more recent one below, is disabled so nobody answers something old.
• A notice in the chat every time their project or their appointment changes.
• Their reference images, which they can attach at any time.

The client never sees anything about your credits, your plan or your settings.

If you run out of credits

The agent hands the conversation to you with its usual message (“I have passed the conversation to <your name>, who will answer you here.”) and you get a notification that says it is because of credits. The client sees nothing about credits. You can reply yourself without limit, and your plan's credits renew every month; how much you have left is in Home → Settings → Plan. If you do not want to wait for the renewal, buy a top-up (extra credits that do not expire) in your account on drop.ink, under Plan; the app does not sell credits. On Trace or Ink you can also turn on automatic top-ups there. When you have credits again, turn the AI switch back on in those conversations or write in them.

Clients who abuse the chat

If someone sends spam, goes off topic, tries to manipulate the agent or insults you, the agent detects it: the first time it is only noted and the conversation goes on; the second time, the conversation is paused for 24 hours, you get a notification and you decide whether to block the client from their profile. A blocked client gets no replies and cannot book. Grave things (a threat, harassment outside the chat, sexual content involving minors) do not wait for the second strike: the conversation is paused the first time and the notification tells you it is grave. These do not count as abuse: saying hello again or asking for a reply while they wait, asking about the price or about you, writing their idea again, intimate areas, dark designs or religious symbols.

You ask the agent for what belongs to your studio: if someone asks it for something else (homework, recipes, translations), it declines in one sentence and goes back to the tattoo, and it never repeats its instructions or what you told it about yourself, even if asked. If the client talks about a health problem (an infection, a fever) or about hurting themselves, the agent answers carefully, tells them to call 112 or see a doctor, without giving medical advice, and hands the conversation to you; the notification does not say why. If the design carries a hate symbol, the agent neither praises it nor offers slots: it tells them that is your call.

What your clients say about their health (an allergy, a medicine, a pregnancy, an illness) does not reach the AI: before the agent reads the conversation, Drop takes those sentences out and leaves a mark that only says there was something about health. You read it as it was written in the chat. The agent does not ask about it, tells the client you will see it and carries on with the tattoo, and you get a notification («Wrote something personal the assistant did not read»), under Messages, so you look at it. If what they say looks like an emergency, the agent still knows it is one: it tells them to call 112 and hands the conversation to you, as above. The filter works by words and may miss one, so do not ask clients for health details in the chat.

If in one day you get many new conversations with no appointment (ten or more, and three times your usual), we notify you, at most once a day: it may be someone spending your credits with several emails. You can block them or switch off AI in new chats.

While the pause lasts, the chat tells you AI paused until the day and time it lifts, and why: · possible spam or, if it was something grave, · grave message (never what they wrote). The card in the list says Paused and the switch appears off. The client sees that you will answer them, and each message of theirs reaches you as a notification. If it was a real client (or your own tests), tap Resume AI or switch the switch on: the pause lifts, the agent answers the waiting message right away and the earlier strikes stop counting.

Frequently asked questions

Can I start a conversation with a client myself?
Yes, if you added them by hand (see Clients): open their chat and write to them. A few minutes later they get an email with the link to your conversation, once they have confirmed their email by entering your chat.

Does the agent give prices?
It never gives or calculates a price. It only repeats the "price from" that the Flash piece they ask about or the project already has; for an idea of their own with no price, it says you will give it. What it does tell them are the payment methods you set in Settings → AI assistant.

How much does each conversation cost?
A typical conversation up to closing a booking uses about 1 credit. No conversation can spend more than 5 credits in one day (12 on Ink).