Privacy Policy

Effective: 2026-10-18

Operator: HigherEdPods Inc., a corporation incorporated in Alberta, Canada, operating intelligence.fm ("we", "us").

This is the beta-period version of our policy. It describes what the service actually does today, including the parts that are unfinished. Counsel will review it before we charge anyone. Where a detail is still being confirmed we say so in brackets rather than guess.

1. Who we are and what this covers

intelligence.fm gives podcasters their analytics inside their own Claude. This policy covers:

It does not cover services you use alongside us and that have their own policies: Anthropic's Claude, Google Chrome and the Claude in Chrome extension, your Google account if you choose to sign in with Google, your password manager, or the podcast platforms themselves (Apple Podcasts Connect, Spotify for Creators, YouTube Studio, Amazon Music for Podcasters). When your Claude reads our answers or drives your browser, Anthropic's terms and privacy policy govern that part.

Questions, requests and complaints go to support@intelligence.fm. A human (one of the two founders) replies. Our privacy officer, for the purposes of Alberta's Personal Information Protection Act (PIPA), is reachable at the same address. Postal address: Edmonton, Alberta; our street address is available on request to support@intelligence.fm.

2. How the service works, in one paragraph

You add our connector to your own Claude and sign in with your email. When you ask Claude to collect your analytics, your Claude drives your own Chrome browser, already logged in to each platform as you, downloads the exports those platforms offer you, and pushes the files to us. We store the files, parse them into metrics, and answer your questions with those metrics and the definitions and caveats that go with them. Separately, our own servers read a small number of public web pages and feeds about your show on a schedule, such as your show's reviews and chart positions on Apple Podcasts; section 3.10 describes exactly what. That reading uses no login of yours and never touches your platform dashboards. We never ask for, receive, store, or see your platform passwords, session cookies, or API secrets. If you opt into one of the optional API routes, the credential lives in your own computer's keychain; we hold only the name of that keychain item and its expiry date so we can remind you before it lapses.

3. What we collect

3.1 Account information

3.2 Your podcast analytics

The exports your Claude pushes to us contain aggregate figures only: plays, streams, listeners, unique and engaged listeners, followers and subscribers, watch time and listening hours, completion and retention curves, and, from YouTube, estimated revenue in US dollars. They are broken down by show, by episode and by day, week or month. They also carry your show and episode titles, release dates, platform identifiers, and the public RSS feed you give us so that episodes can be matched across platforms.

There is no listener-level data in the exports we ingest. No listener names, emails, device identifiers, or locations. (The one place a listener is named anywhere in the service is a public review, under the name its author chose to show publicly; see section 3.10.)

Aggregate audience breakdowns. If you choose to, you can also have your Claude collect the aggregate demographic breakdowns a platform shows you about your audience: how your plays divide by age bracket, by gender, and by country, for your show and for episodes you name, one month or week at a time. Today this comes only from Spotify for Creators, through the optional Spotify API route (section 2: the credential stays in your own computer's keychain), and only when you run that step. What we store is each bucket's figure as the platform serves it (for example "plays by listeners aged 35–44" or "share of plays from Canada") and the total it is a share of. These figures are about groups, never about a person: nothing in them names, identifies or locates an individual listener, and a country share is not a listener's location. Where a platform withholds a bucket because too few people are in it, we record that it was withheld and never estimate it.

Two honest details. First, we keep every export exactly as uploaded, unchanged, for as long as your account exists, and we only parse the columns we recognise; if a platform ever adds a column we do not parse, it still sits in the stored file. Second, identical files are stored once, keyed by their content hash, so a byte-identical export pushed twice does not create two copies.

Everyone who is a member of your workspace can see all of the workspace's shows and analytics; during the beta we add a member only on the account holder's written request.

3.3 Notes you write

You can attach annotations to a show or a date ("we launched mid-roll ads on June 1", "moved hosts", "featured on a chart"). That text is yours, free-form, and we store it as written so it can ride along with future answers whose date range overlaps it.

3.4 The question log

Every call your Claude makes through the connector is recorded, one row per call. Each row holds: which tool was called, for which show, the arguments as sent (with file contents replaced by a byte count, anything that looks like a secret, such as a token, cookie or password, replaced by a placeholder, and long text cut at 1,000 characters), whether we answered or refused and why, how long it took, your intelligence.fm user ID (never your email), the name and version of the client that connected, and a timestamp. Notes you write in an annotation are therefore also visible in the log, up to that length.

Who can read it: members of your own workspace (your own questions) and intelligence.fm staff. What we use it for: understanding what podcasters ask that a dashboard could not answer, fixing slow or broken tools, and deciding what to build next. We do not currently delete question-log rows. We intend to set a retention period and will update this policy when we do.

3.5 Collection telemetry and failure captures

Each collection run records when it started and stopped, why it stopped, what it collected, and any "drift" it hit (a platform changing its page). When a run fails, your Claude captures the page it was on: a copy of the page's HTML, a small JSON note (the page URL, title, first heading and an excerpt of its text), and, if your computer permits it, a screenshot. It uploads these to us so we can fix the problem centrally without asking you to reproduce it. Because these are captures of your logged-in platform dashboard, they can contain whatever that page showed: your account or channel name, an email address in an account menu, episode lists, and figures such as revenue. They are never parsed into metrics and are used only for debugging. They are deleted after the issue is resolved.

A short text alert about the failure (platform, kind of failure, reason) is also posted to our internal Slack. It contains no personal information.

Once a week we also post an internal staff report to the same Slack, summarising how the connector was used: the tools called and how many times, refusal codes, show titles, the first characters of workspace and show IDs, and short samples of the recorded arguments and error messages described in section 3.4. Those samples carry the same redactions the log itself has — file contents are replaced by a byte count and anything that looks like a secret by a placeholder — so the report never contains a file you uploaded or a credential. It is visible only to intelligence.fm staff.

3.6 Optional API-route status

If you opt into an API route for a platform, we record the platform, the name of the keychain item on your machine, the dates it was created and expires, and whether it is active. Never the credential itself. The connector rejects anything that looks like a pasted secret.

3.7 Skill watermark

The collection instructions ("skills") we serve to your Claude carry a watermark: your workspace ID, your user ID, the show they were issued for, when, and a checksum. This lets us attribute a leaked copy to an account. It contains no email address.

3.8 Infrastructure logs

Our hosting providers keep ordinary server logs (IP addresses, request paths, user agents, timestamps) for security and operations. We do not run any third-party analytics, error-tracking or advertising software on the connector.

3.9 Our website

https://www.intelligence.fm uses Fathom Analytics, a privacy-focused, cookieless service that counts visits without identifying visitors. We set no cookies. Joining the waitlist sends an email from your own mail client to our inbox; we keep that email to contact you.

3.10 Public information about your show, read by our servers

Your analytics (section 3.2) arrive from your own Claude. This section is different: our own servers read it, on a schedule, from public web pages and feeds, not from your browser and not under any login of yours. They read only what anyone can see without signing in, only about shows in your workspace, and never a logged-in page or your platform dashboards. Browser collection from your platform dashboards is unchanged and still works exactly as section 2 describes.

What they read:

We keep a copy of each page or feed exactly as our servers received it, the same way we keep your exports (section 3.2), and derive the figures above from it. When a page cannot be read as expected, the run is recorded as failed, with that copy attached so we can fix it.

Reviews contain other people's personal information, so a few commitments apply to them specifically. We read them for one purpose: to show you what listeners have publicly said about your own show. They are visible only to members of your workspace and to intelligence.fm staff, and the text reaches your Claude only when you ask about reviews. We never contact reviewers, never try to identify them, never combine their names with any other source, and never use them for advertising or publish them. Anyone whose review appears here can write to support@intelligence.fm about it, and has the rights set out in sections 10 and 11.

4. What we do not collect

5. How we use your information

We do not sell your information and we do not use it for advertising.

6. AI processing

Two different things happen with AI, and it is worth separating them.

Your Claude. Every question you ask and every answer we return passes through your own Claude session. That is governed by your agreement with Anthropic, not by us. We see only the tool calls your Claude makes to our connector.

Our weekly digest. To write the digest we send a small, structured bundle to Anthropic's API: your show title, the week's dates, per-platform aggregate figures for the week and the week before, the titles of your top episodes, and the caveats that apply. It never includes your raw files, your email address, or any question-log rows. Anthropic processes it under its commercial API terms. If the API is unavailable, we send a plain, non-AI summary instead.

7. Who we share information with

We share information only with the service providers that run the service on our behalf, and only what each one needs:

ProviderWhat they do for usWhat they holdWhere
SupabaseDatabase, sign-in, and file storageEverything in sections 3.1 to 3.7 and 3.10United States (AWS us-west-2)
Fly.ioRuns the connector server and our scheduled jobs, including the reading of public pages in section 3.10Traffic passing through, server logsToronto, Canada
AnthropicGenerates the weekly digest textThe digest bundle in section 6United States
ResendSends our emailYour email address and the email contentsUnited States
SlackInternal failure alerts and our weekly internal staff reportAlert text with no personal information; in the weekly report, tool names and counts, refusal codes, show titles, the first characters of workspace and show IDs, and redacted argument and error samples (never file contents, never credentials)United States
Webflow, Cloudflare, FathomHost and measure our websiteVisit counts; standard web logsUnited States
Brave SearchWeb search, to find your show's public pages on listening apps (section 3.10)Your show's public title, inside a search query limited to one listening app's website, sent from our servers; nothing about you. Brave's search privacy notice says it keeps de-identified query text, including the words searched, and states no retention periodUnited States (Brave's search servers are hosted by AWS in the US, per its search privacy notice; the search API names no region of its own)

Apple, OP3 and the listening apps whose public pages and feeds we read (section 3.10) are not on this list, because they are sources, not service providers: we send them nothing about you, only an ordinary request for a public page or feed, which identifies itself as intelligence.fm. The one detail about your show we send to a source is to OP3: your show's public RSS feed address, so that OP3 can say whether it counts your show. (Brave Search, above, is a provider: it receives your show's public title so that we can find its pages.)

We may also disclose information when the law requires it, to protect the rights or safety of anyone, or as part of a merger, acquisition or sale of the business, in which case we will tell you before your information is transferred under a different policy.

8. Where your information lives

Your account data and analytics are stored in the United States by Supabase. The connector itself runs in Toronto. Because HigherEdPods Inc. is an Alberta company, Alberta's PIPA and Canada's PIPEDA govern how we handle personal information, wherever you are. Under PIPA section 13.1 we are telling you here that we use service providers outside Canada to store and process personal information for the purposes described in this policy, and that you can ask support@intelligence.fm how to obtain more information about our policies and practices concerning those providers. Information held in the United States may be accessible to US authorities under US law.

If you are outside Canada, the same rules apply to us. Local law may also give you rights; see section 10.

9. How long we keep information

After an account lapses (for example, a trial or plan that has ended), we intend to keep the data for 12 months, warn you before deleting it, and then delete it. That mechanism is designed but not yet built; until it is, nothing is deleted automatically and the terms above apply.

Deletion on request: email support@intelligence.fm and we will delete your account, your exports, the metrics derived from them, your annotations, your question-log rows, any failure captures, the public information about your show described in section 3.10, and stored files, within 30 days, and confirm when it is done. There is no self-service deletion button yet; this is a manual process on our side. Backups may retain copies for a further short period before they cycle out.

10. Your choices and rights

11. If you are in the EEA, the UK or California

We are a small Canadian company and most of our users are in Canada and the United States, but we honour these rights for everyone who has them, through the same channel: email support@intelligence.fm.

European Economic Area and United Kingdom. HigherEdPods Inc. is the controller of your personal information. Our legal bases are: performing our contract with you (running the service, sign-in emails, and the digest and reminders about your own account); our legitimate interests (the question log and collection telemetry, to keep the service working and improve it; reading public reviews of your show, which name their authors, so that you can see what listeners have publicly said about your own show, as section 3.10 describes; security; enforcing our terms); and consent, where we ask for it (any future marketing email). You have the rights of access, rectification, erasure, restriction, portability and objection, and the right not to be subject to a decision based solely on automated processing that has legal or similarly significant effects on you, which we do not make. We respond within one month. Your information is transferred to Canada and the United States. Canada's federal private-sector privacy law (PIPEDA), which governs our cross-border handling of your information, is recognised as providing adequate protection by the European Commission and the United Kingdom; for our providers in the United States we rely on their data processing agreements, which incorporate the European Commission's standard contractual clauses and the UK addendum. You may lodge a complaint with your local supervisory authority, or with the Information Commissioner's Office in the UK.

California. We do not meet the thresholds that make the California Consumer Privacy Act apply to a business, and we do not sell personal information or share it for cross-context behavioural advertising. We honour the CCPA rights anyway: to know what we hold about you, to delete it, to correct it, and not to be treated differently for exercising those rights.

12. Security

If a security incident creates a real risk of significant harm to you, we will notify the Alberta Commissioner as PIPA requires and notify you directly.

13. Children

intelligence.fm is a business tool for podcasters. It is not directed at anyone under 18 and we do not knowingly collect information from them.

14. Changes to this policy

We will change this policy as the service changes. The effective date at the top tells you when it last changed. For material changes we will email the address on your account before they take effect. Prior versions are available on request.

15. Contact