RunCell Privacy Policy (SaaS)

1) What this policy covers

This policy explains how RunCell processes data when you: (a) install and use the extension, (b) invoke AI features, and (c) choose to enable or disable cloud sync. This document is intended for our SaaS users on Free, Pro, and Team plans. Enterprise deployments may be governed by separate terms.

2) Data we process

• Account data – email, plan type, license status, basic usage counters.
• Workspace access (local) – the extension can read notebooks, code cells, outputs, and files to assist you. Raw files are not uploaded by default.
• Prompt context – small, relevant snippets extracted from the current session (e.g., a traceback, short code fragment, column names) used to build an LLM prompt.
• Conversations – your messages and model responses in the RunCell chat. If cloud sync is enabled, we store conversation content so you can resume across devices; if disabled, conversation content remains on your device.
• Product analytics & event data – information about how you interact with our UI (e.g., page views, button clicks, navigation flows, performance timings) and basic device/browser metadata. We use this to improve usability and reliability. These analytics events do not include your notebook/file contents or conversation text.
• Feature flags & experiments – non-content data about which features or variants are enabled for your account (e.g., flag keys, variant assignment, exposure timestamps) so we can safely roll out improvements.
• Operational telemetry – minimal, payload-free metrics (timestamps, latency, error codes) used for reliability, abuse prevention, and billing.

3) How AI requests work (data flow)

1. Local scope – You ask a question; the extension gathers only the scoped context (e.g., current cell code or a recent error).
2. Send to cloud – Your question and the minimal context are sent to our cloud prompt orchestrator.
3. Prompt assembly – We construct the final prompt (your query + chosen context + short interaction history) on our server.
4. LLM call – We forward the prompt to the LLM provider you select (e.g., OpenAI, Anthropic, Google, or others available in-product). We instruct providers not to use your data for training where data-control options are available.
5. Response – The model’s output is streamed back to your extension.
6. Storage – If cloud sync is off, we do not persist conversation content server-side; if on, we store it so you can pick up where you left off.

4) Training & model improvement

• Paid plans (Pro/Team) – “Privacy Mode” is on by default: we do not use your prompts, context, or model outputs to train or fine-tune models.
• Free plan – “Privacy Mode” is not available: we may retain and use your prompts, context, and model outputs to improve RunCell, including training or fine- tuning.
• Third-party LLMs – Regardless of plan, we instruct third-party LLM providers not to train on your data where their controls support this. Note: providers may perform limited abuse/quality checks as part of operating their services.

5) Storage, retention & deletion

• Conversation content – Stored only if you enable cloud sync. You can disable sync at any time; when disabled, new conversation content is not uploaded, and previously synced content can be deleted from your account.
• Prompt context – Processed ephemerally for the active request unless cloud sync is on (in which case it may appear within your stored conversation history).
• Operational telemetry – Contains no prompts or file contents; retained for a limited period for reliability, security, and billing.

6) Security (summary)

• Transport & storage – Data sent to our service is protected with TLS; any content we store is encrypted at rest.
• Access control – Role-based, least-privilege access for our staff; access is limited to operating and securing the service.
• Vulnerability management – We maintain standard patching and monitoring practices appropriate for a SaaS in early availability.

7) Your choices & controls

• Choose the model provider from those offered in-product.
• Cloud sync toggle – on/off for storing conversation content for cross-device use.

8) Subprocessors (categories)

We use service providers to deliver RunCell, including:

• Cloud hosting & storage
• LLM providers (to generate AI responses)
• Product analytics (e.g., Microsoft Clarity) to collect interaction events for UI/UX optimization
• Error monitoring & operational analytics (payload-free)
• Billing & payments

We maintain contracts requiring appropriate data protection measures. A current list of subprocessors will be posted on our site and updated as needed.

9) Enterprise note

If your organization requires stricter isolation or self-deployment, contact us. Enterprise deployments may be governed by separate terms and may involve different data flows and controls.

10) Children

RunCell is not directed to children under 13 (or a higher age as defined by local law).

11) Changes to this policy

We may update this policy from time to time. When we do, we will revise the “Effective date” above. Material changes will be communicated via the product or email when appropriate.

12) Contact

Privacy questions: support@runcell.dev