Skip to content

Legal

Privacy Policy

Last updated: November 6, 2026

1. Who We Are

philter.io is operated by Philter Inc., a Delaware corporation (“philter,” “we,” “us,” “our”). We build and run a workspace for each client engagement, where meetings, notes and mail come in and decisions, tasks, deliverables and briefs come out. This policy covers the workspace and the public site at philter.io, including the access-request form: what information we hold, how we use it, and who processes it on our behalf.

2. Information We Collect

Account information

Your name, email address, language, timezone and notification preferences, and your sign-in credentials. Each sign-in is kept with the family of browser and operating system it came from — never the full browser string — and how it was verified. Access is by invitation only; there is no public sign-up. Sign-ins and access to shared material — a share link’s views, and each download of a document, a tool, the org deck or the board — are recorded in an audit log, and so is each call an AI tool you connect to the workspace makes, with the network address and client software it came from.

Content you bring in

Meeting transcripts and recordings (from Fireflies.ai or pasted in), notes, email you forward, copy or send to the workspace and its attachments (documents, images, audio), WhatsApp chat exports, voice notes, files members share with their team in the workspace, a spreadsheet you share with our service account, files behind share links (Dropbox, Google Drive) in the mail and WhatsApp chats you bring in, and material a member’s own AI tool adds. From this material the platform derives decisions, tasks, deliverables and deadlines, which members review, and briefs.

Connected services and forwarded mail

When you connect Fireflies.ai, we access what the API key you provide permits, and we store that key encrypted. When you forward, copy or send mail to the workspace’s intake address, we receive the message and its attachments, including the names and email addresses of its senders and recipients, and we check the message’s authentication before acting on it. Where a message is held for a member’s review, the review card shows the sender’s name and address, the subject, and the recipients’ names and domains — never the recipients’ addresses.

People who are not members

The material your team brings in can name people who are not members — the participants of a meeting, the senders and recipients of mail and chats, the people your company lists in the workspace. We hold that information as part of the material it came in with, and use, keep and delete it as this policy describes for that material.

Usage data and error reports

On the public site, PostHog collects usage data (page views, clicks) under a random identifier held in the page’s memory — it is never stored in your browser, and each page load starts a new one. In the workspace, analytics are tied to your membership identifier — a random id, not your name or email — and analytics store nothing in your browser there either. On both, PostHog receives the IP address each request comes from and derives an approximate location from it; the address itself is discarded, not kept with the event. Because analytics store nothing on your device, we ask no consent for them; we rely on our legitimate interest in understanding how the site and the workspace are used (GDPR Art. 6(1)(f)). Error reports are collected via Sentry with the same identifier and a scrubbed description of the page, never the content on it. We do not sell this data or share it for advertising.

Access requests

When you request access through philter.io, we collect your name, work email and company, and — if you give them — your role, an indication of project scale and any message you write. The request is stored and sent to us by email, and we send at most one acknowledgement to the address you gave; it carries nothing you typed. Nothing is added to a mailing list.

Cookies

In the workspace, strictly necessary cookies keep you signed in. When you sign in, a strictly necessary security cookie gives your browser a random identifier (each of the workspace’s two sign-in pages keeps a copy), so we can tell you when your account is signed in from a browser it doesn’t recognise; we keep only a one-way fingerprint of it with each sign-in, and the cookie lasts up to 400 days, renewed each time you sign in. If you hide the workspace’s navigation, a preference cookie remembers that choice in that browser for a year; it holds nothing else. If you choose your AI assistant on the workspace’s tools guide, another preference cookie remembers which one, in that browser, for a year; it holds nothing else either. Your colour theme is kept in your browser’s local storage, on the public site and in the workspace; it holds nothing else either. Analytics store nothing in your browser — no cookie, no local storage — on the public site or in the workspace, and an identifier an earlier visit left is removed. We set no advertising cookies.

3. How We Use Your Information

  • Provide, maintain and improve the workspace
  • Analyze the material you bring in with AI and propose decisions, tasks, deliverables and deadlines for your team’s review; a few measured, reversible kinds of update the platform applies itself and marks as its own
  • Send briefs, digests and notifications by email, in your language and timezone
  • Keep an auditable record of what was decided, by whom, and from what
  • Answer access requests and, unless you ask us not to, contact requesters about philter
  • Respond to support requests
  • Protect the service, including rate limiting and abuse prevention

4. Data Storage and Security

Your data is stored on Supabase (PostgreSQL and file storage) in the United States. Only our servers — the application and its background worker — and our own operators’ tools reach the database; browsers never reach it directly, though a WhatsApp export or a file shared with your team uploads straight to file storage over a signed link. All data is encrypted in transit (TLS) and at rest, credentials for connected services are stored encrypted, access is tiered per member, and every access to shared material is logged.

5. Data Sharing

We do not sell your data. We share data only with the service providers necessary to operate the platform, each processing it only to deliver its service and under its data processing agreement:

  • Anthropic (AI processing — the runtime provider; processes your content to return a result and does not train on it)
  • OpenAI (AI processing — the failover provider, on the same terms)
  • Supabase (database, file storage and sign-in)
  • Vercel (application hosting)
  • Trigger.dev (background processing — most AI analysis and the scheduled sends run there)
  • Resend (email — sending, and receiving mail sent or forwarded to the intake address)
  • Deepgram (speech-to-text for voice notes and other audio files)
  • Langfuse (AI observability — since October 2, 2026, records only metadata about AI runs — which model ran, timing, token use and cost — never their prompts or outputs)
  • PostHog (product analytics)
  • Google Workspace (our own mail — founders@philter.io, where access requests and your messages to us arrive)
  • Sentry (error monitoring and performance traces)
  • Upstash (rate limiting — request identifiers, IP addresses and, for a trusted sender, its address or domain)

These providers run primarily in the United States. Fireflies.ai, Google Drive and Dropbox are sources you connect or link; we read from them and send them nothing of yours.

6. Data Retention and Deletion

Your data is retained for the duration of your engagement. Items you delete are archived: they leave the working views — an archive view lists them and can restore them — and remain in the workspace’s record so that a decision can be traced to what produced it; a workspace admin can delete an archived item permanently. Permanent deletion of an item, an account or the whole workspace happens on request, and at the end of an engagement as agreed. An access request is kept for as long as we may follow it up; you may ask us to delete it, or to stop contacting you, at any time. To request deletion, contact us at founders@philter.io.

7. Your Rights

You may request access to, correction of, or deletion of your personal data at any time by contacting us. If you are in the EU or EEA, you also have the right to lodge a complaint with your data protection authority. You may end a connected service at any time — by disconnecting it in the workspace where it offers that (the meeting recorder does), revoking its key with the provider, or asking us to remove it — which ends our access to that service’s data.

8. Changes

We may update this policy from time to time. We will notify you of material changes via email or an in-app notice.

November 6, 2026

A correction of fact, checked against how the service works. No right, obligation or commitment changed.

  • Analytics: PostHog no longer keeps the IP address with the event. It still derives an approximate location from the address before discarding it.

November 1, 2026

Corrections of fact, checked against how the service works. No right, obligation or commitment changed.

  • Now listed: the preference cookie that remembers the AI assistant you chose on the workspace’s tools guide.
  • Deleted items stay in each section’s archive, where they can be restored, and an admin can delete an archived item permanently.
  • Besides our application servers, our background worker and our operators’ tools reach the database, and your browser uploads a WhatsApp chat export or a file shared with your team straight to file storage over a signed link.
  • We read share links in the mail and WhatsApp chats you bring in, not links pasted on their own.
  • Briefs are written and sent without a review step, so they are no longer listed among what members review.
  • A held mail’s card shows the sender’s name and address, the subject, and the recipients’ names and domains, never their addresses.
  • Now listed: each sign-in’s browser and operating-system family and how it was verified, files members share with their team, material a member’s own AI tool adds, the colour theme kept in your browser, and the security cookie’s two copies.
  • The service providers’ roles, stated in full: Supabase also runs sign-in; Trigger.dev runs most, not all, AI analysis; Deepgram also transcribes audio files other than voice notes; Langfuse receives metadata about AI runs, never content; Sentry also receives performance traces; Upstash also holds a trusted sender’s address or domain.
  • Access to shared material is spelled out: a share link’s views, and each download of a document, a tool, the org deck or the board.
  • People who are not members: the material your team brings in can name them, and their information is handled like the material it came in with.
  • Analytics: PostHog receives the IP address each request comes from and derives an approximate location from it, so the public site’s analytics are no longer called anonymous.
  • Google Workspace is listed: it hosts our own mail, where access requests and your messages to us arrive.
  • Ending a connection: you can also disconnect the meeting recorder in the workspace.

9. Contact

Questions about this policy? Email us at founders@philter.io.