Skip to content

Privacy Policy

What we collect, why, who processes it, and how long we keep it.

1. Who we are and what this covers

Pixeltable, Inc. ("Pixeltable", "we") is based in San Francisco, California, USA. This policy covers our website, the dashboard, Pixeltable Cloud (hosted databases, services, the Media Store, the API, the pxt CLI's connection to Cloud, and the hosted MCP server), and our free tools and demos (the "Services"). It does not cover the open-source Pixeltable library, which runs on your own machine, or sites we link to, such as GitHub, Discord and docs.pixeltable.com. Your use of the Services is also subject to our Terms of Service.

We are the controller of account, billing, usage, website and communications data. For Customer Content in Cloud we are a processor acting on your instructions. You are the controller of any personal data in it, including data about your own end users.

2. What we collect

From you and your sign-in provider

  • Account and organization: name, email address, profile image, sign-in method (email or a social provider you choose), organization memberships and roles, and sign-in times, through WorkOS AuthKit. API keys are issued by WorkOS; we keep a SHA-256 hash, the name, the permissions and the creator.
  • Billing: Stripe collects your name, email, billing address, tax ID and payment method. We receive a customer ID, plan, subscription and invoice status, and usage totals. We never see or store full card numbers.
  • Customer Content: table data and schemas in your databases, media files in the Media Store, project code you deploy (archives and built images), runtime secrets, service configuration, and the logs your databases and services write. The names of your organization, databases and services appear in the hostnames of your endpoints (for example org-db.svc.pxt.run) and in public DNS.
  • Messages and submissions: emails you send us; pxt feedback reports (your text, the report type, Pixeltable and Python versions, operating system, agent name; no account is needed); waitlist and feedback forms (email, name, job title, answers, page URL). Do not put secrets or personal data in them.
  • Chat and tools: Builder (formerly PixelAssist) sends your message, up to 16 earlier messages and, for a question, the page path to Groq; a request to build an app goes to Anthropic instead when Build mode runs on Anthropic. Ask sends your query. PixelScore sends pasted code, or files fetched from a GitHub repository you name, to Groq. Some free tools and demos (for example the background remover) send what you upload or type to a backend we host on Render, and may forward it to a model provider such as Groq or Google. Other tools run entirely in your browser and send nothing. Do not submit secrets or sensitive personal data to any of them.
  • Careers: applications go to Gusto through our job listings. We do not collect them on this site.

Collected automatically

  • Usage records: compute hours of each database and service, daily storage samples, resource names and states, the changes you submit, and when each database was last used. We use them for billing, plan limits and support.
  • Technical data: IP address, browser and device, URLs, timestamps and status codes in the logs of Vercel, Cloudflare, AWS and our gateway; approximate location from the IP address; and the events listed under Cookies.
  • Campaign attribution: the first time you visit outside the dashboard we store in the pxt_ft cookie the campaign that brought you (UTM source, medium, campaign and content, Google click IDs, the landing path) and your Google Analytics client and session IDs. When you sign up we copy it to your account. We then send sign-up, organization-created, MCP-connected and qualified-sign-up events to Google Analytics from our servers, with your user ID, that attribution, and a tier based on whether you created an organization, whether your email is from a free provider or a company domain, and what you have done since. We use it to measure which campaigns bring accounts and to optimize our Google advertising.
  • Trials: a trial made with pxt new or by an AI agent has no account. We store a keyed hash of the caller's IP address, not the address, to enforce limits, until the trial is claimed or deleted.

3. How we use it

PurposeDataBasis (EEA/UK)
Run accounts and Cloud; support youAccount, usage, Customer ContentContract
Bill, collect tax, prevent payment fraudBilling, usageContract; legal obligation
Secure the Services, limit abuse, debugTechnical data, trial IP hash, logsLegitimate interests
Service, usage and billing emailsName, email, usageContract
First email about an hour after sign-up, and one email to accounts inactive for 15 daysName, emailLegitimate interests
Answer chat, tool and feedback requestsWhat you submitContract; legitimate interests
Measure campaigns and optimize our advertisingAttribution, user ID, email domain type, account activityLegitimate interests
Understand and improve the websiteAnalytics, replaysLegitimate interests
Comply with law, enforce our TermsAny of the aboveLegal obligation; legitimate interests

We use Customer Content only to provide, secure and support Cloud for you, including debugging, abuse prevention and legal compliance. We do not use it to train machine-learning models, ours or anyone else's, unless you agree in writing. We make no decisions about you by automated means that have legal or similarly significant effects. Where we rely on consent, you can withdraw it at any time.

4. Who we share it with

We do not sell personal information. The website loads no advertising or retargeting tags, but the events described under campaign attribution go to Google. We share data with the vendors below, which process it for us, and with others only as this section says.

VendorWhat it does for usData
Amazon Web ServicesControl plane; the Kubernetes clusters that run your databases and services; project archives and build contexts (S3); built images (ECR); logs (CloudWatch). Region us-east-1.Customer Content, account and usage records, logs
PlanetScaleThe Postgres database behind each Pixeltable database, and the control-plane database. AWS us-east.Customer Content (table data), organization and account metadata, usage records
CloudflareDNS for pxt.run; R2 storage for the Media Store; the Worker that serves media through signed links.Media files, hostnames, request IP addresses
WorkOSSign-in (AuthKit), organizations and members, API keys, Vault for runtime secrets, agent registration for trials.Account data, API key records, runtime secrets
StripePayments, subscriptions, invoices, usage metering, tax.Billing details, usage totals
VercelHosts the website and the dashboard backend; Speed Insights.Request data, session cookies, performance metrics
ResendSends our email.Name, email address, message content
GroqAnswers Builder, Ask and PixelScore requests, and powers some free tools.Messages, page path, submitted code
AnthropicWrites the app for a Builder request to build one, when Build mode runs on Anthropic.Messages
AirtableStores waitlist and feedback-form submissions.Email, name, job title, answers, page URL
SlackReceives pxt feedback reports.Report text, Pixeltable and Python versions, operating system
Google (Analytics, Ads and Gemini)Website analytics, measuring which campaigns bring sign-ups, and Gemini-powered image generation on Pixelimage.Usage events, device data, approximate location, user ID, campaign attribution, sign-up tier. Pixelimage: the images and prompts you submit for generation.
Microsoft ClaritySession replays and heatmaps on every page except the dashboard and claim pages.Page interactions, device data
RenderHosts the backends of some free tools and demos.What you submit to those tools
GustoReceives applications from our careers listings.Application materials

We also disclose information where we believe the law requires it, to enforce our Terms, to protect the rights, property or safety of Pixeltable or others, and in a merger, financing, reorganization or sale of assets. Services and providers you connect yourself, such as model APIs you call from your code, your own storage bucket, social sign-in providers and AI assistants connected through MCP, receive data under your agreement with them, not ours.

5. Cookies and browser storage

NamePurposeLifetime
wos-sessionRequired. Keeps you signed in (sealed WorkOS session).Up to 7 days
pxt-help-quotaCounts guest questions to Builder.24 hours
pxt-oauth-state, pxt-oauth-attrRequired. Protect the sign-in redirect and carry campaign attribution through it. Deleted when sign-in completes.10 minutes
pxt_ftFirst-touch campaign attribution (Section 2). Not set on the dashboard.180 days
WorkOSSign-in pages set their own cookies.WorkOS sets it
StripePayment and fraud prevention on checkout and billing pages.Stripe sets it
Google Analytics (_ga, _ga_*)Website analytics.Up to 2 years
Microsoft Clarity (_clck, _clsk)Session replays and heatmaps. Not loaded on dashboard or claim pages, and stops when you move to one.Up to 1 year

Your browser's local storage holds the organization you selected, the page to return to after sign-in and sidebar layout. Session storage holds your Builder conversation until you close the tab. Vercel Speed Insights measures page performance without cookies.

We do not show a cookie banner, so Google Analytics, Clarity, Speed Insights and the first-touch cookie load without asking first. In production, Google Analytics runs across the site and Clarity runs on every page except the dashboard and claim pages. You can block either in your browser or with an extension, and the site works without them. We do not respond to Do Not Track signals.

6. How long we keep it

DataRetention
Account and organization dataWhile the account exists, then as long as law, tax, security or a dispute requires.
Billing and usage recordsAs long as tax and accounting law requires, including after the account is deleted.
Customer ContentUntil you delete it. Deleting a database, a service or an organization starts teardown at once and is permanent: we keep no copy to restore. Backups kept by our providers expire on their own schedules. A stopped database or service keeps its data, and so does one stopped by a plan's usage limit, by Community's idle stop or by the end of Pro.
Cached mediaAfter a media object or database is overwritten or deleted, a holder of a signed link may still get the old copy for up to two hours: up to one hour from Cloudflare's cache, then up to one more hour in their browser.
Database and service logs7 days. The dashboard and the CLI read the last 1 day on Community and 7 days on Pro and Enterprise.
Unclaimed trialsDeleted with everything in them 48 hours after creation. A record that a person claimed a trial (user ID, organization ID, time) is kept to enforce the claim limit.
Campaign attribution180 days in your browser; on your account, with its event flags, while the account exists.
Messages to us and feedbackAs long as needed to answer, fix or improve, then deleted.
AnalyticsAs long as the retention set in our Google Analytics and Clarity accounts.

We do not store Builder conversations in a database; Groq and Anthropic process them under their own terms. When we no longer need personal information, we delete or de-identify it.

7. Where it is processed

We are in the United States, and Cloud runs in AWS us-east-1 and PlanetScale on AWS us-east. Cloudflare stores Media Store files in R2; we set no storage location, so Cloudflare chooses it, and its network serves them from many countries. Vercel and Cloudflare serve the website and media from locations near the visitor. For transfers from the EEA, UK or Switzerland we rely on the Standard Contractual Clauses (and the UK addendum) in our vendors' data processing terms.

8. Security

We protect data with TLS in transit, encryption at rest provided by our infrastructure providers, access controls and tenant isolation, as described on our Security page. No method of transmission or storage is completely secure.

9. Your rights and choices

Depending on where you live, you may have the right to access, correct, delete, export, restrict or object to our processing of your personal information, and to withdraw consent. To use a right over data we control, write to [email protected]. We may need to verify your identity, and we answer within the time the law sets (30 days in the EEA and UK, 45 days in California). If you are in the EEA or UK, you may complain to your data protection authority.

Organization administrators can delete databases, services, media, API keys and the organization in the dashboard or with the pxt CLI, and can read and export their data with Pixeltable's own tools. If you are an end user of a Pixeltable customer, ask that customer; we will help them respond. For Customer Content we act on the customer's instructions and refer requests to them.

To stop welcome and product email, reply to it or write to [email protected]. Our emails have no unsubscribe link yet. You will still receive account, billing, usage and security messages.

10. California (CCPA/CPRA)

In the last 12 months we collected identifiers (name, email, account IDs, IP address), commercial information (plan, billing and usage records), internet activity (logs, analytics), and professional information you give us (company, role). We collect it from you, your devices and the vendors above, and use it for the purposes in Section 3. We do not sell personal information. Sending the events described in Section 2 to Google may count as "sharing" for cross-context behavioral advertising under California law; to opt out, email the address below and we will stop sending your events. We do not use sensitive personal information to infer characteristics about you. You may know, access, correct and delete your personal information, and we will not treat you worse for asking. To use these rights, email [email protected]; an authorized agent may act for you as the law permits.

11. Children

The Services are not directed to children under 13, and we do not knowingly collect their personal information. If you think a child has given us any, contact us and we will delete it.

12. Changes

We may update this policy and will change the date below. For material changes to how we treat personal information we control, we will also email the address on your account, or notify you in the Services.

13. Contact

Pixeltable, Inc., San Francisco, California, USA. Email [email protected] or use our contact page. A data processing agreement is available to Enterprise customers on request.

Last updated: November 17, 2026