Privacy Policy

Information we collect

We collect account and workspace information, contact and billing details, product usage data, support messages, and the content and settings you provide to run your workflows.

Connected accounts and project memory

When you connect a social account or community — a Telegram user account, a Telegram Business account, a Telegram bot, or VK — we record that account's activity into a journal scoped to your project. That includes messages and their later edits and deletions, posts and comments, reactions, membership and other service events, presence, and profile and chat information for the people and chats the account sees.

Collection is forward-only. It starts at the moment you connect an account: we record what happens from then on, and we do not import anything from before the connection.

Files and media

Attachments a connected account can download — photos, video, voice and audio, documents, stickers, and profile photos — are downloaded and stored encrypted. Each one is described or transcribed by an AI model so its content can be searched as text. Office documents and archives are read for the text they contain.

Other people in your chats

A connected chat necessarily includes the other people who write there. Their profile information and the content they send are stored under your project alongside your own. You are responsible for having a lawful basis to connect an account and to process the third-party information it exposes.

How we use information

We use information to provide the Sylog Crew application, authenticate users, operate AI workflows, publish or queue approved social content, improve reliability, prevent abuse, and communicate about the product. Project memory exists for one purpose: so the agents you run can act with context on your behalf.

AI and automated processing

Stored content is sent to our model provider, Google Vertex AI, to describe and transcribe media, to compute the embeddings that make memory searchable, and to answer the questions you ask about your own memory. Vertex AI processes this content under enterprise terms and does not use it to train its models.

Storage and encryption

Media bytes and the verbatim copy of what a provider sent are sealed with an application-owned key held in Google Cloud KMS, bound to the organization and project that own the record, so a stored record cannot be opened outside the workspace it belongs to. The database is encrypted at rest. Every memory read and write carries the owning organization and project, so one customer's memory is never reachable from another customer's workspace.

Retention

By default we keep project memory for as long as the project exists, so search and automation stay available. You can set a retention window per project, after which entries older than that window are redacted.

Your controls

The memory section of a project lets you browse everything that is stored. From there you can choose, chat by chat, whether a chat is remembered at all — turning one off drops the memory already stored for it; set a cap on what memory processing may spend; delete any individual entry; erase a specific person entirely, choosing whether that also blocks us from ever collecting them again or leaves them collectable if they write to you later; export everything, including the stored files; and delete the project.

Deletions on the source platform

We mirror what happens where the content was originally sent. When a message is deleted on the platform it came from, we redact it in our record and remove it from the search index. When a message is edited, we record the edit and treat the newest version as the current one.

Deleting a project or an organization deletes its stored content and permanently destroys the associated account credentials.

Consent

Connecting a personal account is an explicit step. Before that flow starts we show what the project will collect and require you to confirm it.

What we do not do

We do not sell personal data. We do not use customer content to train models. We do not mix data between customers: every read and write is scoped to the organization and project that own it.

Service providers

Sylog Crew uses cloud infrastructure, authentication, database, email, analytics, and AI providers to run the product. These providers process information only as needed to provide their services to Sylog Crew.

Contact

For privacy questions, contact us at contact@sylog.app.