Privacy Policy
Effective October 15, 2026
Dolora is a personal AI assistant provided by CYFRA LLC, doing business as MakeBox AI. This policy covers the Dolora website, workspace, and Chrome extension. Contact sales@makebox.ai for privacy questions or requests.
You choose what Dolora receives. Page tools read the current page or your selected text when you use them. Saved memory starts disabled. Dolora does not monitor your browsing in the background or sell your data for advertising.
Information we process
- Account information: your email, name, user identifier, authentication session, and preferences. Google sign-in shares basic identity information through Supabase Auth; it does not connect your Gmail inbox.
- Content you submit: prompts, selected webpage text, the page title and URL, conversation context, and files or images you attach for a task. A page can contain personal or confidential information; review what you choose to share.
- Your workspace: complete saved chat messages and answers, chat titles and model choices, project folders, generated media, mindmaps, artifacts and their versions, comparisons, manually saved memories, and facts learned from conversations when you enable automatic memory. Memory source references and processing status link learned facts to the conversation that produced them.
- Billing information: plan, credit usage, subscription and transaction identifiers, and payment status. Stripe processes payment details. Dolora does not store full payment-card numbers.
- Operational information: request identifiers, model usage and cost, error codes, IP addresses and necessary server logs used for security, reliability, and abuse prevention.
How information is used
We use this information to authenticate you, perform the task you request, save and retrieve your work, calculate credits, manage subscriptions, provide support, and protect the service. Your content is not used for advertising or sold to data brokers. We do not use it to train our own general-purpose AI models.
AI processing and service providers
Requests go through MakeBox servers and OpenRouter to the model provider serving the model you choose. Providers represented in Dolora may include OpenAI, Anthropic, Google, Perplexity, xAI, Black Forest Labs, Runway, Kuaishou/Kling, MiniMax, and ByteDance. Search may use Exa through OpenRouter. PDF processing may use the chosen model provider or Cloudflare through OpenRouter. These services receive the content needed to complete the request and process it under their applicable policies. Provider availability and retention practices can differ.
When you enable Dolora memory, relevant saved facts may be included in new AI requests. Automatic memory can send limited extracts of newly completed conversations, memory-search queries, and relevant facts through OpenRouter for AI processing. The models used to organize and find memories may differ from the model selected in your chat. Personal and project memories are stored as part of your Dolora account on infrastructure we operate. This feature involves the external AI processing described here.
Supabase provides authentication and database infrastructure. Google Workspace delivers account emails from sales@makebox.ai. Stripe processes subscriptions and payments. Our hosting provider, Contabo, hosts the Dolora service and generated media. We share only information needed for these functions, or as required by law and for legitimate security purposes.
Review OpenRouter’s privacy policy, Supabase’s privacy policy, and Stripe’s privacy policy.
Social accounts, posts, and messages
Dolora Social connects publishing accounts and Instagram/Facebook messaging only when you choose to set them up. HighLevel provides the social connection, scheduling, and messaging infrastructure. A separate workspace is mapped to each Dolora account. Publishing permissions and messaging permissions are separate; connecting one does not automatically enable the other. OAuth credentials stay on our servers.
We process the social account details, post drafts, selected media, conversation history, and recipient information needed for these features. Posts and replies are sent only after your explicit confirmation in Dolora. Text, hashtags, images, and reply drafts are generated through Dolora's AI service providers described above; this release does not use HighLevel AI for generation. Selecting AI reply drafting shares the relevant conversation text with those providers.
Opening first-time social connection setup may create a restricted connection profile using your verified Dolora email. Before continuing, Dolora asks for your consent and explains that the provider may send a setup email and require you to finish sign-in in the secure setup window. No profile password is returned to Dolora's client. Media chosen for publishing is made available to the publishing provider for delivery and may become public when the post is published. Disconnecting Dolora does not delete posts or messages already delivered to a social network. Review HighLevel's privacy policy and the policies of the social networks you connect.
Chrome permissions and page content
The extension uses a side panel for its interface, local storage for preferences and sign-in sessions, identity for Google sign-in, context menus for selected-text actions, and activeTab with scripting for page content you request. It connects to Dolora’s API and Supabase Auth. It does not request access to your browser history or read every tab automatically. The URL and title of a page you submit are part of that request’s context.
Dolora’s use of information received from Google APIs follows the Chrome Web Store User Data Policy, including its Limited Use requirements. Human access to content is limited to support you authorize, security needs, legal obligations, or permitted aggregated operations. We do not transfer this data for advertising, credit decisions, or general-purpose model training.
Saved conversations and projects
Full chat messages are saved on our servers in your account, including when long-term memory is off. You can reopen conversations, change the selected model, export a chat as Markdown or JSON, and archive or restore chats and projects. A bounded recent history or a shorter summary may be sent to a model to continue a conversation; creating a summary does not erase the original saved messages. Archiving keeps the underlying data so you can recover it.
Accounts have separate data and memory. Personal memory can be used across your models and new conversations. Project memory is used only for conversations in that project; personal memory may also be available there. Switching models does not create a new memory account. Per-chat and project memory controls can further restrict use of that context.
Memory choices and automatic learning
The master memory setting and Automatic memory start switched off for new accounts. Existing accounts keep their current settings, which are shown on the Memory page. Long-term memory is separate from your saved chat history. You can manually save a preference or fact, choose personal or project scope, edit it, move it to Trash, or restore it. Manual memories in Trash are excluded from new AI requests.
Automatic learning works only while both the master memory setting and Automatic memory are enabled and the applicable chat or project allows memory. It processes new completed conversation turns after you enable it; this release does not automatically backfill your historical chats. Useful facts are extracted and retained in the account or project memory space that applies to that conversation. AI-generated facts can be incomplete or incorrect, so you can inspect the learned facts and their source conversation.
In Memory, open a learned source and use Forget learned facts to remove that source and its derived facts from Dolora memory. The original saved chat remains in your account. Independently saved manual memories and facts learned from other sources are separate. Source status, request identifiers, usage records, and necessary deletion records may remain without the forgotten source text.
Turning memory off excludes saved personal and project memory from new requests and stops new automatic learning. Turning Automatic memory off stops new learning while saved facts can still be used if the master memory setting remains on. These switches do not erase existing saved chats or facts and cannot reverse processing that has already started. Queued, unstarted learning work is cancelled when either setting is turned off. An in-progress or uncertain source may need to finish or be checked before it can be forgotten.
Storage, retention, and deletion requests
Sessions and preferences are stored on your device. Prompts, complete chat messages, results, generated files, project folders, creations and their versions, and manual memories are stored in your Dolora workspace. Learned facts and their source references are stored in Dolora memory on our servers. Temporary learning-queue text is cleared after processing or cancellation; this does not erase the original conversation. Original PDF attachments are processed for the requested task and are not added as original files to the creations library. AI providers may retain request data under their own applicable policies.
We keep workspace information while it is needed to provide your account and saved history. Trashed or archived items remain stored so they can be recovered. You may request access, correction, export, or permanent deletion by contacting sales@makebox.ai from your account email. We verify requests before acting. Some billing, security, and legal records may need to be retained. Deleting Dolora data does not automatically delete data belonging to another MakeBox product or copies retained by an AI provider under its policy.
Security and international processing
We use HTTPS, server-side provider credentials, and account-scoped database access. Generated artifacts are previewed in an isolated sandbox without access to the parent workspace, extension APIs, or network. No service can promise absolute security. Information may be processed outside your country by the providers listed above.
Cookies and analytics
Dolora uses necessary browser storage for authentication and preferences. This release does not include advertising trackers, session replay, or third-party marketing analytics in the Dolora workspace or landing page.
Children and changes
Dolora is intended for adults aged 18 and over. If a child has provided personal information, contact us to request its removal. We may update this policy when the service changes and will indicate the effective date. Material changes will be communicated through the service or account email where appropriate.