Holler is a SaaS product, so there is more to disclose than a typical desktop tool. Here is the full picture in plain language: what we collect, who we share it with, and how to make it all go away.
The short version
What we collect
Holler uses Clerk for authentication. When you sign up, Clerk stores your email address, the name you provided, and any OAuth identifiers (for example, your Google or LinkedIn account ID) if you used social sign-in. Holler reads that profile back to know who you are when you log in.
Clerk is a third-party identity service. Their own privacy policy applies to the data they hold on our behalf. See clerk.com/privacy.
When you paste a URL or upload a document (Word or PDF), Holler fetches or extracts the text and sends it to Anthropic or Google Gemini for processing. Once the transformation finishes, the source article is discarded from Holler's servers. It is not retained, indexed, or used to train any model.
What IS retained is the output: the platform-ready posts and images Holler generated for you. Those live in your History so you can come back to them.
When you train Brand Voice on 3 to 5 URLs of your published writing, Holler extracts characteristics of your voice: sentence rhythm, vocabulary patterns, level of formality, humor frequency, paragraph length, and similar features. Those characteristics are saved as a profile tied to your account.
The original text of the articles you point Brand Voice at is not stored. Only the derived features are. Think of it as a fingerprint of how you write, not a copy of what you wrote.
There are no third-party analytics scripts on the Holler app (no Google Analytics, no PostHog, no Vercel Analytics, no Mixpanel). The only usage data Holler collects is the operational kind needed to run the service: how many credits you have used, how many transformations you have started, and the timestamps of those events. That information lives in your account record and is used to bill credits and show your History.
The marketing pages at holler.skytech.io (the homepage, this privacy page, the help page) similarly have no third-party analytics on them today.
Sub-processors
The services Holler relies on to deliver the product:
No. Holler uses the commercial API tiers of both providers, which contractually prohibit using customer content to train the underlying models. Your source articles and the generated outputs are not added to any training corpus.
Each provider publishes its own data-handling policies for commercial API customers. You can review them directly: Anthropic's commercial terms and Google Gemini API terms.
Stripe handles the entire payment flow. Your card number, expiration, and CVC are entered into a Stripe-hosted form and never touch Holler's servers. We receive only a tokenized reference and the metadata needed to grant your credits (amount paid, success/failure, plan tier).
Stripe's own privacy policy applies to the financial data they hold. See stripe.com/privacy.
Resend is the transactional email service Holler uses for system messages: signup confirmation, credit purchase receipts, and the occasional account-related notification. We do not use Resend to send marketing email. If you want product updates, that is a separate opt-in via the AI for the Rest of Us newsletter at stevechazin.com.
Supabase and Vercel host our infrastructure in US data centers. AI processing through Anthropic and Google may route through their respective US-based infrastructure. If you are an EU or UK user, your data may be transferred to and processed in the United States under standard contractual clauses with each sub-processor.
Your rights
Email steve@skytech.io from the address tied to your Holler account and ask for a full deletion. Within 30 days we will remove your Clerk account, your Brand Voice profile, and your generated content History. Operational records that we are legally required to retain (such as Stripe transaction records for tax purposes) may persist according to the retention rules of those services.
We are working on a self-serve deletion button in Settings. Until that ships, the email path is the canonical way.
Yes. Email steve@skytech.io and request a data export. We will send a JSON file containing your Brand Voice profile and your full History (the inputs and outputs of every transformation you have run) within 30 days.
If you are protected under GDPR, CCPA, or a similar regime, you have the right to access, correct, port, and delete the data Holler holds about you. Email steve@skytech.io with your request and we will respond within 30 days.
We do not sell personal information under the CCPA definition. We do not run third-party advertising on the Holler app.
For most cases, yes. Source articles you submit are not retained in Holler's database. Anthropic and Google's commercial APIs do not train on your content. Your generated output and Brand Voice profile live only in your account, never shared with other users.
If you are working with material under a strict NDA or regulatory obligation (HIPAA, attorney-client privileged work, classified or export-controlled content), check with your compliance team before using any third-party AI tool, including Holler. Holler is not currently positioned for those use cases.
Yes, over time. When Holler adds a new feature that changes the data picture (a new sub-processor, a new data type), this page will be updated and the change will be announced on the AI for the Rest of Us newsletter at stevechazin.com. We will never quietly expand data collection.
Holler is built by Steve Chazin, a one-person operation, under the skytech.io umbrella. There is no parent company harvesting your data on the side. steve@skytech.io reaches a human directly.
Last updated: May 2026.