Parmenion subprocessors
Last updated: October 21, 2026
Parmenion is run by CHIBBS MANAGEMENT LLC ("we", "us"). To run it, we use other companies that handle data for us. For your brand's data (its ad data, uploaded reviews, goal, winner rules, strategies, board and uploaded files, and Parmenion AI history), we act for your brand, so these companies are our subprocessors. For your account, billing and usage data, we decide how it's used, and they act as our service providers. This page lists both: what we use each one for, what data it gets, where it keeps or processes that data, and where to read its terms.
This page goes with our privacy policy, our terms of service and our How to delete your data page. Our data processing addendum (DPA) is part of our terms and uses this page as its list of subprocessors. For a signed copy, email sam@chibbsmanagement.io.
Where things stand today
- Parmenion isn't open yet. As of the date above, no company below holds data from customers' brands, except the one test ad account described next. Stripe holds what founding members give at its checkout.
- Our test builds use a made-up brand with made-up numbers, plus one real ad account, connected with its owner's written consent to test our Meta connection and for Meta's review of our app. Vercel already hosts those private test builds, and Neon already hosts their test database.
- Vercel also hosts our public website, getparmenion.com. It has no analytics or tracking tags, and any free tools on it send nothing you type in them. Its home page has one box to join our list: the email address you give there, and two optional answers, go from your browser straight to Loops.
- Google Workspace already hosts our email inbox, sam@chibbsmanagement.io, so anything you email us is stored there.
- Loops is in use for that list. The other companies are marked "Planned". When one starts handling real customer data, we'll mark it "In use" here, with the date.
- Reading the comments on your ads and your Facebook Pages and Instagram accounts, keeping copies of your ad images and of still frames from your video ads, the board in the Sandbox, uploading files to it, Parmenion AI on the board, the analyzer, the briefer, the ad library and the free trial are still being built. What this page says about them applies when Parmenion offers them.
The data we mean
The list below uses these names:
- Account data: your name, work email, team, role and sign-in history, the email addresses of people you invite, and, while Parmenion is in beta, the email addresses we invite to test it.
- Set-up answers: your revenue range, new ads a week and how you heard about us, each optional.
- Brand settings: your brand's name, website and country, your growth goal and winner rules, which campaigns test new ads, which sale counts, labels your team adds to ads, a profile of your customers if you fill one in (when Parmenion offers it, your customer map's avatars and pains), your strategies, and, when Parmenion offers them, your logo and colours, the angles you keep and concepts you write, what you tell us you sell, subscriber counts and a Super Strategy onboarding's answers. Apart from the labels, they're kept while your brand is on a plan, and deleted as our privacy policy says. The labels are kept and deleted with your ad data.
- Ad data: what we read from Meta with the access you give us. We only read it. That's your ad accounts, campaigns, ad sets, ads, their budgets and daily results, and your ad account's activity history; ad text, which images and videos your ads use, and the IDs of the Facebook or Instagram posts they run as, and, when Parmenion offers them, copies of your ad images and of still frames from your video ads; when Parmenion offers them, the names and IDs of the Facebook Pages and Instagram accounts you share, and the text of comments on your ads; and the ID Meta gives your business. It also covers what we work out from them, such as winners, labels, summaries, scores and your plan, and, when Parmenion offers the analyzer, its reads, the transcripts we make of video ads and the numbers we measure from their sound (never the sound itself). We drop commenters' names, and keep their IDs only as a code made with our secret key. Without our key, no one can link the code to a person.
- Uploaded reviews: review text, rating, date and product, with reviewers' names and emails removed.
- Your ad library: when Parmenion offers it, the ads your team uploads to its library (JPG, PNG, WebP, GIF, MP4 and MOV, or a ZIP), with location data removed.
- Board content and uploaded files: when Parmenion offers the board, on plans that include it, the notes, text, docs and flowcharts your team puts on your brand's board in the Sandbox, and the files it uploads there: CSV, PDF, JPG and PNG, kept as uploaded. Files you export from Meta yourself, such as Ads Manager reports, are kept and deleted with your board, and count as data from Meta when we report a security incident to Meta.
- Meta connection tokens: the keys that let us read your ad data. We store them encrypted.
- Billing data: your billing contact, business name and billing address, your phone number, and the brand details and Discord name its checkout asks for, if you join founding access, any tax ID you enter, your plan, and invoices, which name your brand; and, to give each business one free trial, your website and email domains and Stripe's code for your card. We never see your full card number.
- Trial reminder data: for our email 24 hours before a free trial's first charge, the owner's email address, your brand's name, its plan and price, and when the trial ends. No card digits and nothing about ads.
- Usage data: which pages you open, what you click, how far you scroll, the type and size of files added to or removed from your board, how Parmenion AI is used (which features, how often, when they fail, thumbs up or down, token counts and what each use costs us, never what's typed to it or what it writes), feedback you give us and answers to our in-app survey, why you cancel, your country (the IP address is used for it, then discarded), reports of errors you run into and our servers' logs, and our own records of payments (plan, amount and our ID for your brand).
- Security data: IP addresses, browser details, the bot check's signals from your browser when you sign up, and our log of important actions, such as connecting, exporting and deleting.
- AI request data: what we send an AI model to do one job for your brand, and what comes back. That's the numbers our code worked out, your ads' text, images and video frames where a task needs them (and, when Parmenion offers the analyzer, their on-screen text and transcripts), what your team types for the briefer, the text of comments from Meta and of reviews uploaded as reviews (never names or IDs), questions you type in the Ask box, the board items and uploaded files a task needs when Parmenion offers AI on your board and you use it there (files go with any names in them), and the labels, summaries, answers, board drafts, reads, angles and concepts the AI writes back.
- Email data: the email addresses we write to (yours, those of people you invite, and those of testers we invite), the emails we send, whether you've unsubscribed from our marketing emails, and, for marketing emails, your name, plan and sign-up date. For people who join our list on our website: the email address, the date they joined, and, if they give them, whether they run a brand or work as a strategist and their monthly ad spend range.
- Support data: what you write to us by email.
We never send an AI model your Meta connection tokens, your billing data, or commenters' names or IDs from Meta.
The list
| Subprocessor | What we use it for | Data | Where | Status |
|---|---|---|---|---|
| Vercel | Hosting for Parmenion and our public website, background jobs, the bot check on sign-up, file storage and logs | All the data above except support data passes through it. It stores, when Parmenion offers them, copies of your ad images and video frames, the files you upload (to the board and the ad library), job records and logs | United States (Washington, D.C.) for the app, files and jobs. Its global network delivers pages and files, and may cache them briefly in a country near you | Planned for customer data. Already hosts our private test builds, which hold made-up data and the one test ad account, and our public website |
| Neon (contract with Databricks, Inc.) | Our main database | The data above, except support data, sent emails, and, when Parmenion offers them, the copies of your ad images and video frames and the files you upload, which sit in Vercel's file store. It also holds the rows of CSVs you upload. Tokens are stored encrypted | United States (AWS, N. Virginia) | Planned for customer data. Already hosts our test database, which holds made-up data and the one test ad account |
| Anthropic | Claude labels your ads, sorts comments and reviews by the questions customers ask, writes the short summaries around the numbers our code works out, answers Ask questions, helps draft and explain strategies, and, when Parmenion offers AI on your board, reads the files you upload and drafts items for your board. When Parmenion offers them, it also reads your ads for the analyzer, finds angles and writes concepts, and drafts avatars from comments | AI request data | United States, and Anthropic may process AI requests in other countries | Planned |
| Stripe | Checkout, subscriptions, free trials, invoices, tax, the billing portal, and most of our billing emails: receipts, failed-payment notices, the yearly renewal reminder and its reminder when a free trial starts | Billing data, plus the payment details you enter on Stripe's page | United States | In use for founding access, from October 20, 2026 |
| PostHog | Product analytics, error reports and server logs, heatmaps, feature rollouts, our in-app survey, and Parmenion AI's counts and cost | Usage data, tied to an account ID in the app. No ad data, tokens, emails or Parmenion AI text | United States (AWS, N. Virginia) | Planned |
| Loops | Sends our own emails: sign-in links, team invites, beta invites and required notices, such as changes to our terms or this list, and security incidents, and one email before we close an inactive account. It also sends our one billing email, the reminder 24 hours before a free trial's first charge, and our marketing emails, such as launches, guides and offers, each with a one-click unsubscribe. It keeps the list people join on our website | Email data and trial reminder data | United States | In use, for the list on our website. Planned for the rest |
| Google Workspace | Our business inbox for support, privacy, refund and security emails, and our internal records | Support data, and internal records, which never hold ad data | Google's data centers, which may be in the United States or other countries | In use for our inbox |
Their terms and security pages
| Subprocessor | Agreement and DPA | Their own subprocessors | Security |
|---|---|---|---|
| Vercel | DPA | List | Security |
| Neon | Platform terms · DPA | In the platform terms | Security · Trust center |
| Anthropic | Commercial Terms · DPA | List | Trust portal |
| Stripe | DPA · Privacy policy | List | Security |
| PostHog | DPA | List | Security · Trust center |
| Loops | Terms · DPA · Privacy policy | List | In its DPA |
| Google Workspace | Data processing addendum | List | Security |
More about each one
Vercel
- Our app code, file store and background jobs run in Vercel's Washington, D.C. region.
- Pages, file copies and the files you upload reach you through Vercel's global network. Copies of your images and files can stay in its cache near you for up to 1 hour, including after we delete them.
- Vercel's DPA says it processes data mainly in the US, and may transfer it to other countries under standard contractual clauses.
- When Parmenion offers them, copies of your ad images and video frames, and the files you upload, sit in a private store that needs a key for every read.
- Background-job records are kept for 7 days after each job. App logs are kept for up to 30 days.
- Vercel also keeps its own platform backups for 30 days, and we can't reach them. If they include our file store, a deleted image, video frame or uploaded file can stay there for up to 30 days.
Neon
- Neon hosts our Postgres database on AWS in the US. The region is fixed when the database is created.
- We use Neon through Vercel. The contract is with Databricks, Inc., Neon's parent company, and its DPA applies.
- Neon encrypts data at rest and in transit.
- Deleted data leaves Neon's restore history within 7 days.
Anthropic
- We call Claude directly, through Anthropic's API, under our own Commercial Terms and DPA with Anthropic. Our code works out every number, and Claude writes the words around them. It never gets direct access to our database.
- Anthropic's Commercial Terms say it may not train models on what we send.
- Every request that carries your data is covered by zero data retention. Anthropic then doesn't keep the request or the answer once it has replied, except where the law requires it or a request is flagged for misuse.
- When Parmenion offers AI on your board, we send the files it needs inside each request, under the same terms and zero data retention, and never through Anthropic's file storage.
Stripe
- You enter your payment details (a card, Apple Pay or Google Pay) on Stripe's own pages, so they go straight to Stripe. We never see your full card number.
- For sales tax, Stripe collects your billing address and, if Stripe supports tax IDs for your country, your tax ID.
- Stripe acts for us when it handles your payments. It also uses some payment data for its own purposes, such as preventing fraud and meeting its legal duties. For those uses it's an independent controller under its own privacy policy.
- Stripe's privacy policy also lists training AI models to power its services and protect against fraud. It doesn't say whether that covers data it handles for us.
- Stripe sends most of our billing emails: receipts, failed-payment notices, the yearly renewal reminder and, when a free trial starts, a reminder with a link to cancel. It keeps its log of them for 60 days. The first charge after a free trial carries Stripe's note that the trial has ended.
- Stripe gets no ad data.
PostHog
- We use PostHog's US Cloud.
- We tie usage data to an account ID only, not to your name or email. PostHog uses your IP address to work out your country, then discards it.
- PostHog gets no ad data. It records that you clicked, not the text or attributes of what you clicked. Page addresses and titles carry no ad names or Meta IDs, and error reports are cleaned of ad data before they're sent.
- Screen recordings are off.
- For Parmenion AI, PostHog gets the feature, model, token counts and cost, never what was asked or written. For payments, it gets our own record of the plan, the amount and our ID for your brand, never names or card details.
- Events are kept up to 7 years, which PostHog doesn't let us shorten; server logs 14 days. When you close your account, we delete you, with your events, in PostHog.
- We turn off PostHog's AI features and its setting that lets PostHog train its own models on anonymized data.
Loops
- Loops sends every email we send ourselves, from "Parmenion" at hello@getparmenion.com, through Amazon SES, Amazon Web Services' email service. Billing emails come from Stripe, except our reminder 24 hours before a free trial's first charge, which Loops sends with your brand's name, its plan and price, and when the trial ends.
- Loops' own subprocessors list says they process data in the United States.
- How long Loops keeps emails and delivery logs is set by its own terms.
- An invite never adds anyone to our marketing emails.
- When you join our list on getparmenion.com, your browser sends your email address straight to Loops; our website doesn't keep it. If you answer the two optional questions, they go to Loops the same way, with your email address.
- Our emails carry no ad data. Marketing emails are never built from your brand's ad data, and each one has a one-click unsubscribe.
Google Workspace
- It hosts our business inbox, sam@chibbsmanagement.io, so what you email us is stored there.
- We may also keep internal records there, such as our records of security incidents. They never hold ad data.
Not on this list
- Meta. Meta is where your ad data comes from, not a company that works for us. We ask Meta for one permission,
ads_read, and use it only to read, never to change, hide or delete anything. Meta's login also addsemailandpublic_profileon its own, and we don't use them. When Parmenion offers comments, Facebook Pages or Instagram, we'll ask for the extra read permissions those need. Meta sees the requests we make. - Other AI providers. We don't make images or video. We send your ad data, uploaded reviews, board, uploaded files and Ask questions to no AI provider other than Anthropic. We make the analyzer's transcripts ourselves, with no transcription company.
- GitHub. It holds our code and the made-up sample data. It never gets customer data.
- Amazon Web Services' key service (AWS KMS), planned. It will hold the master key that locks the keys we use to encrypt your Meta tokens. It never sees the tokens or your ad data.
- Our vendors' own subprocessors. Each company above uses its own subprocessors, such as cloud hosts. Their lists are linked above.
If you're outside the US
Our providers store your data mainly in the US. Our database, file store and background jobs run in US regions, and Anthropic may process AI requests in other countries. Some subprocessors also handle data elsewhere, as their DPAs allow: Vercel may move data to other countries and caches pages and files briefly near wherever you open Parmenion; Vercel's and PostHog's networks pass requests through servers near you; and Google Workspace may store our inbox in Google's data centers outside the US.
For now, Parmenion is only for businesses based in the United States, apart from founding members. If you're in the EU, the UK or Switzerland, your data is transferred to the US. For your account data, see our privacy policy. Our vendors' DPAs, linked above, say how each one protects data transferred out of those places.
How we tell you about changes
- Before we add a subprocessor, or use one for a new kind of data, we:
- add it to this page, with the date, and note it under "Changes to this page";
- email the owner of each team at least 30 days before it starts handling your data.
- If you object on reasonable data protection grounds, email sam@chibbsmanagement.io. We'll look for a fix with you. If we can't find one, you can cancel that brand's plan, and we'll refund what you prepaid for the time left on a plan paid by the month or year. A lifetime founding seat isn't refunded (section 13 of our terms).
- In an emergency, such as replacing a subprocessor to keep Parmenion running or secure, or because Meta tells us to stop using one, we'll tell you as soon as we can. You can still object, and the same remedy applies.
- When we stop using a subprocessor, we remove it from this page, stop sending it data, and ask it in writing to delete your data. We keep its reply. Stripe keeps the payment records its own legal duties need, under its own privacy policy.
- Our vendors' own subprocessors aren't covered by our 30-day notice. Each vendor announces its changes on its own list, linked above, with the notice its DPA gives us.
Questions
Email sam@chibbsmanagement.io, or write to CHIBBS MANAGEMENT LLC, 1640 Marcie Court, Deerfield, Illinois 60015, United States.
Changes to this page
| Date | Change |
|---|---|
| October 1, 2026 | First version. |
| October 20, 2026 | Board content, uploaded files and Parmenion AI on the board added, ahead of their launch, to the opening, "Where things stand today", "The data we mean", Vercel's, Neon's and Anthropic's rows, the Vercel and Anthropic sections, and "Not on this list". How Parmenion AI is used added to usage data. |
| October 20, 2026 | The free trial: Stripe's trial reminder, and Loops sends our reminder before a trial's first charge, with new trial reminder data. Billing data adds what we check for one trial per business. |
| October 20, 2026 | PostHog's full use: error reports and server logs, heatmaps, feature rollouts, our survey, Parmenion AI's counts and cost, payment records and the country. |
| October 20, 2026 | Ahead of their launch: the analyzer (transcripts and on-screen text to Anthropic), the briefer, the customer map's avatars, the ad library's files at Vercel and Neon, and the brand's look. |
| October 21, 2026 | Loops in use for the list people can join on our website: the email address given there, and two optional answers. The opening, email data, and Loops' row and section say so. The opening no longer names a free tool. |
| October 20, 2026 | The Meta connection: the ad data we read now names which images and videos your ads use and the IDs of the posts they run as, and says that copies of your ad images and video frames, comments, and your Pages and Instagram accounts come when Parmenion offers them. "Not on this list" says Meta is asked for one read permission, ads_read. |
| October 20, 2026 | Stripe is in use for founding access, with the phone number its checkout asks for. Vercel and Neon's test builds also hold one real ad account, connected with its owner's written consent. |