Privacy Policy
This policy explains what personal data Sollar processes, why, who we share it with, how long we keep it and how you can exercise your rights. It applies to both editions of the Sollar apps for Android and iOS (the store app and Sollar Dev), to web access and to the sollar.com, sollar.cn and account.sollar.com websites.
Summary
- Sollar is a messenger for businesses and people, built on the open Matrix protocol. It is in testing (beta).
- There are two editions. The Sollar store app (Google Play and the App Store, including TestFlight) is only the messenger: it has no artificial intelligence (AI) features and does not show conversations with AI agents. Sollar Dev, a test edition installed outside the stores, also has AI agents, transcription, translation, dubbing and other features under test.
- AI features work only for accounts with AI access turned on. Every new account starts without it, and it can only be turned on in Sollar Dev, with your consent. Accounts that already existed before this rule kept access turned on.
- Conversations are end-to-end encrypted. There are exceptions, explained below; the ones involving AI exist only for accounts with AI access turned on, because AI agents do not take part in conversations with anyone who has not turned that access on.
- The app shows no ads, uses no advertising or usage analytics tools, and we do not sell personal data.
- Sollar is intended for people aged 18 or older.
- You can request deletion of your account at any time, as explained in the How to delete your account section.
Sollar editions and AI access
Sollar has two editions, built from the same code and connected to the same server. The same account works in both.
- Sollar store app: distributed through Google Play and the Apple App Store, including TestFlight. It is a messenger, with conversations, groups, photos, files, voice messages, locations you choose to send, contact requests, reporting and blocking. It has no AI features: the code for those features is not in this app.
- Sollar Dev: the test edition. On Android it is a separate app (com.sollar.dev), installed directly from the app.sollar.com website, outside the stores; the team also uses development builds. Besides the messenger, it has the features under test: AI agents, message transcription and translation, dubbing, voice print, reading messages aloud and miniapps.
AI access
- AI features work only for accounts with AI access turned on. Every new account starts without it.
- Access is turned on in Sollar Dev, after you read and accept the AI use notice (version of September 26, 2026). Accounts that already existed when AI access started being controlled per account were migrated with access turned on, without going through the notice, because until then AI features applied to every account. You can turn access off at any time in the Me tab of Sollar Dev, under AI access (AI features on Android), or ask for it to be turned off by writing to kameda@kamedatec.com. Turning it off applies to AI use from then on and does not by itself delete what is already stored, such as your voice print and agent data, which follow the retention rules in this policy.
- Without AI access, agents do not appear in search, contact requests to an agent are refused and the server refuses transcription, translation, dubbing and voice requests. Agents also do not take part in conversations with anyone who does not have AI access: an agent declines an invitation to a conversation or group that includes someone without that access, and leaves a group when someone without that access is in it, joins it or is invited to it. If a message of yours reaches an agent before it leaves, the agent discards it before storing it, logging it or sending it to any AI model.
- The store app does not show conversations with agents or invitations from agents, and you cannot chat with agents through it, even if your account has AI access turned on. Those conversations remain available in Sollar Dev.
Who is responsible for your data
The controller, meaning the company responsible for your data, depends on where you use Sollar:
- In mainland China: 珠海市科铼博科技有限公司 (Zhuhai Colabtec Technology Co., Ltd.), unified social credit code 91440403MAEPAF0A2E, Zhuhai, Guangdong, China.
- In Brazil and all other countries: Colabtec Digital LTDA, CNPJ 65.787.255/0001-91, Alameda Rio Negro, 503, sala 2020, Alphaville, Barueri, SP, 06454-000, Brazil.
For any privacy matter, including contact with our data protection officer, write to kameda@kamedatec.com.
Data we process
Data you provide
- Account: mobile number or email address, password, Sollar ID, display name, profile photo (optional) and preferred language. The password is stored only as a hash, which cannot be turned back into the password.
- Profile and contacts: optional profile details, such as status and region, and the contact requests you send or receive, with their introduction message.
- AI access: whether it is turned on for your account and which version of the AI use notice you accepted, with the date.
- Content: messages, photos, files, voice messages and locations you choose to send.
- Voice print (optional, Sollar Dev only): the sample of your voice you record and the reference text you read, used to create a voice profile.
- Audio you ask us to dub (Sollar Dev only): a clip of the audio may be used to create a temporary profile of the speaker’s voice, as explained in the Voice and dubbing section.
- Companies: if you create or join a company on Sollar, the company name, members, roles and invitations.
- Reports: the identifier of the reported message, conversation or person, the reason given and, if you choose to include it, the excerpt of the message you selected, up to 500 characters.
- Support: whatever you write to us by email.
- Website contact form: name, company, email and message.
Data generated by use
- Technical data: IP address, access dates and times, device model and operating system, app version, session and device identifiers, and error logs.
- Conversation metadata: who takes part in each conversation, group names and activity times. The server needs this data to deliver messages.
- Notifications: if you turn alerts on, a device identifier issued by Google or Apple.
- AI agents (Sollar Dev only, with AI access turned on): instructions, tasks, reminders and requests you create with an agent, and what the agent keeps in memory from its conversations with you.
- Website: aggregate visit counts per page and language, without storing IP addresses. The website uses no advertising or tracking cookies; the only cookie is the team admin session.
What we do not collect
- We do not access your device address book.
- We do not use your location in the background. It is read only when you choose to send it in a conversation.
- The microphone is used only when you record a voice message or, in Sollar Dev, your voice print.
- In Sollar Dev for Android, installed directly from the Sollar website, outside app stores, the app may read an incoming SMS while the code screen is open, only to fill in the verification code. The SMS text is neither stored nor sent.
End-to-end encryption and its exceptions
Sollar conversations use the open Matrix protocol, with end-to-end encryption turned on by default. Messages, photos, files and audio are encrypted on your device and can only be read by the devices of the people in the conversation. Our server stores this encrypted content and cannot read it.
There are important exceptions:
- AI agents (accounts with AI access turned on). Agents are participants. When you talk to an agent or add one to a group, it receives and reads messages like any other participant. To reply, the agent sends the necessary text and images to the AI model providers, and audio to our processing server for transcription, all listed in this policy. Those conversations appear only in Sollar Dev.
- Groups with agents. An agent stays only in groups where every person has AI access turned on. In those groups, your device encrypts messages for the agent too, as it does for any participant, and the agent decrypts them on our server and handles them as described in the previous item. If a group has, or comes to have, someone without AI access, the agent declines the invitation or leaves the group, and any message from that person that reaches it before it leaves is discarded in the memory of our server before being stored, logged or sent to any AI model.
- Transcription, translation, dubbing and reading aloud (Sollar Dev, accounts with AI access turned on). When you request these features, or turn them on in settings, the app sends the audio or text to our processing server without end-to-end encryption. The result comes back to your device. Reading aloud can also happen on the device itself. For dubbing, a clip of the audio may be used to create a temporary profile of the speaker’s voice.
- Voice print (Sollar Dev). The sample you record and the reference text are stored on the Sollar servers in Germany. The voice profile created from them is kept on our processing server.
- Contact requests. The introduction message sent with a contact request is stored on our server without end-to-end encryption, so the other person can read it before accepting. It is kept for as long as the accounts of both the sender and the recipient exist.
- Metadata. Participants, group names, times, profile details and company data are not covered by end-to-end encryption.
- Reports. When you report something, the identifier of the content and the reason you give are sent to our team. If you choose to include the selected excerpt of the message, up to 500 characters, that excerpt leaves end-to-end encryption and becomes readable by our team.
- Notifications. Alerts delivered by Google or Apple include neither names nor conversation content.
Protection also depends on your device. Anyone with access to it while unlocked can see your conversations.
Voice and dubbing
The features in this section exist only in Sollar Dev and work only for accounts with AI access turned on. The store app does not have them.
We treat voice data as sensitive data. It is used only to produce dubbing and is never used to identify or authenticate anyone.
Voice print
- It is optional. You record a sample of your voice reading a reference text.
- The sample and the text are stored on the Sollar servers in Germany (Hetzner) and are included in backups. The voice profile created from them is kept on our processing server in mainland China.
- The profile is used to dub your voice messages with a voice similar to yours, when you ask for it or when you authorize another person in the conversation to dub one of your audio messages.
- Legal basis: your consent. To withdraw it, delete the voice print in the app. The sample, the text and the profile are deleted when you delete the voice print or your account.
Dubbing with the speaker’s voice
- When you ask to dub an audio message that has no confirmed and authorized voice print for it, our processing server creates a temporary voice profile from a clip of about 8 seconds of that audio. This profile reproduces the voice of the person who spoke, who may be you or another person in the conversation.
- The clip and the temporary profile are processed on our processing server in mainland China. The profile is deleted as soon as the dubbing is finished. If processing is interrupted midway, any profile left behind is deleted when the service restarts.
- When you ask for dubbing, or turn it on for the audio you receive, the app asks you to confirm that you have permission from the person who recorded it to process the audio and reproduce their voice. Do not ask for dubbing without that permission.
- Legal basis: the consent of the person who recorded the audio, which you confirm you have obtained. That person can exercise the rights described in this policy by writing to kameda@kamedatec.com.
Why we use data and on what legal basis
| Purpose | Data | Legal basis (LGPD and GDPR) |
|---|---|---|
| Create and maintain your account, deliver messages and sync your devices | Account, encrypted content, metadata and technical data | Performance of the contract (Terms of Use) |
| Confirm your identity with codes sent by email or, for mainland China numbers, by SMS | Email or mobile number | Performance of the contract |
| Record whether your account has AI access and which notice you accepted | AI access status, version and date of the accepted notice | Performance of the contract and compliance with legal obligations |
| Sollar Dev AI features, for accounts with AI access turned on: agent replies, transcription, translation, dubbing, reading aloud and web searches run by agents | Messages received by agents, audio and text sent for processing | Consent, which you give when you turn on AI access and can withdraw |
| Voice print and dubbing with a voice similar to yours | Voice sample, reference text and voice profile | Consent, which you withdraw by deleting the voice print |
| Dubbing an audio message with the speaker’s voice, when there is no authorized voice print | A clip of about 8 seconds of the audio and a temporary profile of the speaker’s voice | Consent of the person who recorded the audio, which you confirm you have obtained before asking for dubbing |
| Alerts about new messages | Device notification identifier | Consent (you turn it on and can turn it off) |
| Security, fraud and abuse prevention, fixing faults and reviewing reports | Technical data, metadata and reports | Legitimate interests and compliance with legal obligations |
| Support and handling of requests about your rights | Contact details and messages exchanged with us | Performance of the contract and compliance with legal obligations |
| Replying to the website contact form | Name, company, email and message | Consent |
| Complying with laws and orders from authorities | The data required in each case | Compliance with legal obligations |
For people using Sollar in mainland China, processing is based on consent or on what is necessary to perform the contract, under the Personal Information Protection Law (PIPL). Where the PIPL requires separate consent, such as for sensitive data and for transferring personal information outside China, processing depends on that consent.
We treat voice data as sensitive data. The Voice and dubbing section explains whose voice each item is, where it is processed and how long it is kept.
International data transfers
The Sollar servers are in Germany. The processing server and some providers are in mainland China, and others are in the United States. Your data may therefore be processed outside the country where you live:
- account, profile, encrypted conversations and, in Sollar Dev, the voice print sample and text, in Germany;
- emails with codes sent through Alibaba Cloud and SMS with codes to mainland China numbers, in mainland China;
- in Sollar Dev, for accounts with AI access turned on: messages received by agents, audio and text sent for processing, voice profiles and temporary profiles created for dubbing, in mainland China;
- emails with codes sent through Resend and notification alerts, in the United States, and, in Sollar Dev, for accounts with AI access turned on, searches run by agents and the text of tasks an agent carries out with Anthropic.
We make these transfers to provide the service you use. We are putting in place with each provider the mechanisms required by applicable law, such as standard contractual clauses. You can ask for information about the safeguards for each transfer, and exercise your rights with these recipients, by writing to kameda@kamedatec.com.
How long we keep data
- Account, profile, content and company data: while your account is active.
- After a confirmed deletion request: we complete deletion within 30 days, including backups.
- Messages you have already sent stay with the people who received them, in the conversation history. Files stored on our server are deleted with the account and no longer open, but they remain on devices where they were already downloaded. We cannot delete those copies.
- AI access: the access status and the version and date of the accepted notice are kept while the account exists.
- Agent data (Sollar Dev): instructions, tasks, reminders and what the agent keeps in memory from its conversations with you are kept while the account exists and are deleted when the account is deleted.
- Voice print: the sample, the reference text and the profile are kept until you delete the voice print in the app or delete your account.
- Temporary voice profile created for dubbing: deleted when the dubbing is finished. If processing is interrupted, it is deleted when the service restarts.
- Audio and text sent for transcription, translation, dubbing or reading aloud: used to produce the result and not kept after processing. Temporary files are deleted automatically.
- Introduction message of a contact request: for as long as the accounts of both the sender and the recipient exist.
- Verification codes: valid for up to 10 minutes and usable only once.
- Technical logs: for as long as needed for security, abuse prevention and fault investigation, or for the period required by law.
- Website contact form: 180 days.
- Records of deletion and rights requests: for as long as needed to show that the request was handled.
- Sollar ID reservation: after the account is deleted, the Sollar ID stays reserved so that no other account can use it.
- Backups: we make one a day, and each is deleted within 16 days. Deleted data may remain in them until they expire, so the periods in this section may extend by up to 16 days. The 30 day period for account deletion already includes backups.
Your rights
At any time, you can:
- confirm whether we process your data and access it;
- correct incomplete, inaccurate or outdated data;
- ask for anonymization, blocking or deletion of data that is unnecessary, excessive or processed unlawfully;
- ask for portability of your data;
- find out who we share your data with;
- withdraw consent and learn the consequences of not giving it;
- object to processing and ask for review of decisions made solely by automated means;
- ask for your account to be deleted.
Brazil (LGPD): you can also file a complaint with the National Data Protection Authority (ANPD).
European Union (GDPR): you have the rights of access, rectification, erasure, restriction, portability and objection, and you can complain to the data protection authority in your country.
Mainland China (PIPL): you have the rights to know about and decide on processing, restrict or refuse it, consult, copy, correct, complete and delete your data, ask for an explanation of the processing rules and withdraw consent. Close relatives of a deceased person may exercise these rights as provided by law.
To exercise a right, write to kameda@kamedatec.com. We reply within 15 days. We may ask for information to confirm that the request is yours. These requests are free of charge.
How to delete your account
For now, account deletion is requested by email:
- Write to kameda@kamedatec.com with the subject “Delete account”, from the email linked to your account, or include the linked mobile number and your Sollar ID.
- The privacy team of the company responsible for your account records the request with the date it was received and replies to confirm it. We may ask you to confirm your identity, for example with a code sent to the linked contact.
- We complete deletion within 30 days after the request is confirmed, including backups.
Requests through the app and through the account.sollar.com/en/account/deletion/ page are not available yet. That page explains what is deleted, what may be kept and for how long.
Minors
Sollar is intended for people aged 18 or older. We do not knowingly collect data from minors. If we learn that an account belongs to someone under 18, we may close it and delete the data. If you know of such a case, write to kameda@kamedatec.com.
Advertising, usage analytics and sale of data
No edition of Sollar shows ads or includes third-party advertising or usage analytics tools. We do not sell personal data and we do not use the content of your conversations for advertising. The use of conversations to train AI is covered in the next section.
Conversations and AI training
- Content from accounts without AI access is never used to train, fine-tune or evaluate AI models.
- Accounts with AI access: we will only use your conversations to train, fine-tune or evaluate AI models if you give specific consent for it, separate from the AI access consent, which you can withdraw at any time. Today this consent is not requested from anyone outside the Sollar team.
- How it works today: the Sollar team manually exports conversations from internal test rooms, where only team members and agents take part, decrypting them on our server with the keys of an agent in the room. The team then selects and reviews excerpts to build training and evaluation examples, which stay in the team’s internal systems. Other people’s conversations are not part of this process.
- The AI providers listed in this policy receive the content needed to produce each reply. What they may do with that content follows each provider’s contracts and policies.
Security
We use end-to-end encryption for conversations, encrypted connections (HTTPS), passwords stored only as hashes, restricted access to servers and daily backups. No system is completely secure. If a security incident may create a relevant risk or harm, we will notify you and the competent authorities, as the law requires.
Changes to this policy
We may update this policy. The version in force is always on this page, with its effective date. If a change is significant, we will tell you in advance, in the app or by email.
Contact
Email: kameda@kamedatec.com
Colabtec Digital LTDA, Alameda Rio Negro, 503, sala 2020, Alphaville, Barueri, SP, 06454-000, Brazil.
珠海市科铼博科技有限公司 (Zhuhai Colabtec Technology Co., Ltd.), Zhuhai, Guangdong, China.
This policy is published in Portuguese, English, Spanish and Chinese. If the versions differ, the Portuguese version prevails for Colabtec Digital LTDA and the Chinese version prevails for 珠海市科铼博科技有限公司.