Sparklets AI Processing Notice.

This AI Processing Notice explains how Sparklets uses third-party AI and service providers to understand your requests, generate outputs, use connected powers, coordinate tasks, process voice or meeting features, and operate the assistant. It is designed to be read together with our Privacy Policy, Subprocessors page, and Terms of Service.

Depending on your request and enabled powers, Sparklets may send the following categories of data to AI and service providers:

• Your messages, prompts, task instructions, approvals, and outputs.
• Preferences, selected memories, profile context, timezone, and other assistant context relevant to your request.
• Connected email, calendar, contact, file, LinkedIn, messaging, or browser content that you authorize Emma to use.
• Call audio, speech audio, transcripts, call metadata, meeting notes, speaker labels, and generated speech where voice features are used.
• Search queries, URLs, business names, locations, addresses, ticker symbols, or other task-derived data used for research, maps, browsing, ride links, or market-data features.
• Operational metadata needed for security, abuse prevention, debugging, routing, billing, and reliability.

Sparklets uses provider categories and named examples including Google Gemini, OpenAI, Anthropic, Deepgram, Cartesia, Hume, ElevenLabs, LiveKit, Twilio, Supabase, Clerk, Vercel, Google Cloud, Stripe, Google Workspace, Microsoft, Apple iCloud, LinkedIn, Telegram, WhatsApp, Browserbase, Firecrawl, Tavily, SerpAPI, Google Maps, Duffel, FlightAware, WeatherAPI.com, MET Norway, Uber deep links, Zapier where enabled, Finnhub, Alpaca, Twelve Data, and Financial Modeling Prep.

Not every provider receives data for every user or every request. Providers receive data only when needed to deliver an enabled feature, connected power, assistant request, background task, support function, security function, or billing and infrastructure function.

Sparklets sends data to AI and service providers to:

• Understand and respond to your requests.
• Plan, execute, monitor, and complete tasks you delegate to Emma.
• Use powers you connect, such as email, calendar, contacts, files, phone, voice, messaging, web research, browser sessions, and connected apps.
• Transcribe audio, generate spoken responses, place or monitor calls, and summarize meetings.
• Maintain security, prevent abuse, troubleshoot errors, operate infrastructure, provide support, and process billing.

Sparklets does not sell your data, use it for ads, or use your private content to train a Sparklets foundation model. Sparklets uses paid, commercial, or production API routes for core AI model processing where available. For those routes, we use provider data-processing terms, contracts, or equivalent safeguards and configure no-training or limited-retention controls where supported by the relevant provider and production tier.

For example, business/API terms for core model providers commonly distinguish paid or commercial API processing from consumer or unpaid product use. Those terms may state that API inputs and outputs are not used for model training by default, are not used to improve products, or may not be used to train provider models. Sparklets relies on those provider commitments where applicable, but provider retention, safety review, abuse monitoring, legal compliance, and deletion practices still vary by provider, endpoint, and account configuration.

Provider retention, review, abuse-monitoring, safety, debugging, legal, and deletion practices vary by provider. Sparklets does not represent that every provider has identical retention rules or that providers never retain data for security, abuse prevention, legal, or service-operation purposes.

Before certain content is sent through AI model paths, Sparklets applies deterministic hard-secret masking designed to reduce exposure of payment card numbers, Social Security numbers, bank or routing numbers, passwords, and API tokens. Sparklets may still process working personal information, connected-service content, files, audio, transcripts, memories, and task context when needed to provide requested assistant features.

Sparklets requires affirmative AI processing consent before assistant use. If this Notice materially changes, Sparklets may require renewed consent before additional assistant processing.

You can limit or withdraw future processing by:

• Not connecting optional powers.
• Disconnecting connected services or revoking provider permissions.
• Turning off mobile permissions such as microphone, contacts, photos, camera, or location.
• Reviewing, deleting, or adjusting category permissions for memories where product controls support it.
• Exporting data, deleting data, deleting your account, or contacting privacy@sparklets.ai.

Sparklets' memory-governance controls determine whether memory categories may be used directly in assistant context, shown only as limited descriptors, or withheld from assistant context. Confidential memory is handled with stricter controls and is not exposed as ordinary assistant context.

Withdrawal does not undo processing that already occurred, but it can limit future processing. Some features may stop working if you withdraw consent, disconnect powers, or remove context that Emma needs to perform a requested task.

Sparklets performs tasks based on your instructions, permissions, connected powers, approvals, and automation settings. You remain responsible for your instructions, authority to use connected data, approval of important actions, compliance with law and third-party terms, and decisions or commitments made based on Sparklets outputs.

Questions about AI processing or privacy controls can be sent to privacy@sparklets.ai.