Privacy Policy

Who we are and who decides

Lucelo (formerly Luciola) is a service for building chat assistants in Telegram and VK. It is run by an independent developer; there is no company yet.

For your account data, we decide how it is used. Your customers’ conversations belong to you: you are responsible for them, and we process them only on your behalf to run your assistant.

If your assistant collects leads, you decide which data to ask for, and you are responsible for making sure customers understand why you ask for it and how you will contact them. We process this data only on your behalf.

What we collect

• Account. Email and a hashed password. If you sign in with Google: name, email and photo. With Telegram: name, username and photo (no email or phone). With VK: VK ID, name and photo.
• Workspace. Assistants, scenarios, reminders, settings and connected channels. Bot and community access tokens stay on our servers and never reach the browser.
• Your customers’ conversations. Messages people send to your assistants and the replies, plus the name, username and ID that Telegram or VK provides.
• Leads. If your assistant collects leads, we store your customers’ answers: name, phone and any other fields you set (for example, a service, a time or an address). A customer shares the phone number from their Telegram profile only themselves, with the “Share number” button; they can also type a number by hand. Answers to an unfinished lead are stored too, until the customer finishes it or starts over.
• Materials for building an assistant. Photos, PDFs and links you upload so Studio can recognise prices, hours and an address. Content is sent for recognition; we save confirmed assistant facts on the server, not an archive of the source files.
• Notifications. Your Telegram user ID, if you connect Telegram to receive reminders and notifications. In the new-lead notification we show the customer’s answers without the phone number: a notification may be visible on a locked screen.
• Technical logs. Time, errors, request details and AI usage counters. Logs do not contain message text.
• Cookies. Only necessary cookies: one keeps you signed in, the others remember interface settings such as language and the sidebar. No analytics or advertising cookies.

Why we use it

• run your account, workspace and assistants;
• sign you in and protect the service from abuse;
• receive and send messages in the channels you connect;
• collect your customers’ leads and tell you about new leads;
• send you reminders and notify you when a conversation waits for a person;
• answer your requests.

Who we share data with

We do not sell your data or your customers’ data and do not use it for advertising. We share it only with the subprocessors below, without which the service cannot work, and when the law requires us to.

Today the database is in the EU (Frankfurt).

Other subprocessors, such as hosting, sign-in and AI, work in other countries, and data passes through them. So we do not promise that all data stays in one country.

AI features

AI features are optional. Today there are three: understanding customer messages in bots, when a customer writes in their own words instead of pressing a button; reply drafts and translation in Dialogues; and recognising menus and price lists from files you upload when building an assistant. Message understanding is on by default; you can turn it off for each assistant with the “Understand questions in their own words” switch on the “Channels and publishing” page. Reply drafts are on for every assistant and are prepared only when a customer is waiting for a person; you can turn them off in Dialogues with the “Suggest replies” switch. Recognition runs only when you upload a file yourself.

To understand a message, the model receives the customer’s text, limited conversation history, the current step and relevant assistant facts. Order parsing receives item names and variants, quantities and notes. Reply drafts use recent messages; translation uses the selected message. Separate phone and address fields from the order are not included in the order-parsing request. Contacts or other personal data typed into ordinary messages may still be included: complete anonymisation is not guaranteed.

Requests pass through OpenRouter, Inc. to a model hosting provider. The default models are DeepSeek, Qwen and MiMo; a model’s author may differ from the company processing the request. OpenRouter selects an endpoint for each call; providers offering these models include DeepInfra, CoreWeave, Together and others. Our requests require no collection for training and an endpoint with zero retention of request and response content after processing (ZDR). If none qualify, we do not remove these restrictions and the request may fail.

These restrictions apply to model endpoints. OpenRouter has separate content-logging settings and retains technical metadata such as time, model, token counts, cost and errors. Its rules and retention are described in OpenRouter’s Privacy Policy; we do not promise zero retention of all metadata. Translations and drafts are stored in your workspace. For material recognition, OpenRouter and the model receive file content, such as PDF page images or a page’s text. We save confirmed assistant facts, not a server-side archive of source uploads.

Retention and deletion

We keep data while your account exists. There is no automatic deletion period for conversations yet.

Leads are stored together with the conversation while your account exists and are deleted with it. There is no automatic deletion period for leads yet. A single lead can be deleted on request by email, just like a single conversation.

You can delete your account in Studio settings. This immediately deletes your workspace, assistants, conversations, leads and reminders and cannot be undone.

When you delete an assistant, it disappears from Studio and its bots stop answering at once. You can restore it for 30 days. We keep its data — conversations, orders and storefront photos — for those 30 days to protect against abuse, even if you remove it from the trash; then we erase it automatically and irreversibly. We delete the assistant’s topic in the Lucelo bot together with its messages when you remove it from the trash or when those 30 days are over.

Studio cannot yet delete a single conversation or a single customer’s data. Send the request to admin@lucelo.app and we will do it by hand within 30 days. Database backups may keep a copy for a short time until they are overwritten.

Security

Connections use HTTPS. Database rules let each account see only its own workspace. Passwords are stored hashed. Bot tokens and keys are kept only on the server. If a breach affects your data, we will tell you without undue delay.

Your rights

You can ask to see, correct, export or delete your data, or object to how we use it: write to admin@lucelo.app. We reply within 30 days.

If you are a customer of a business that uses Lucelo, contact that business first. It decides about your data, and we help it answer you.

Children

Studio accounts are for adults. We do not knowingly create accounts for children. Your assistant’s audience is your responsibility as the owner.

Changes

When the policy changes, we update the date at the top. We announce important changes in Studio or by email before they apply.

Contact

Questions and requests: admin@lucelo.app.