Privacy Policy

Last updated: 16/04/2026

Welcome to Stacked ("we," "us," or "our").

This Privacy Policy explains what information we collect, how we use it, and your choices.

We've written this in plain English; no hidden surprises.

If you have questions, contact us at info@stacked.coach.

0. Data Controller

The data controller responsible for the processing of your personal data under the GDPR is:

• Legal entity: Stacked AI

• Address: Theresienstr. 18 90403 Nürnberg

• Contact email: info@stacked.coach

For data-protection inquiries, please use the contact email above.

1. Your Account

To use Stacked Coach, you must create an account. We use Replit as our authentication provider.

When you sign up, we receive:

• Your email address

• Your first and last name

• A unique account ID provided by Replit

You do not create a password with us, authentication is handled by Replit.

Upon creating your account, you receive free credits managed on our servers.

We use this information to:

• Save your coaching sessions

• Let you access your history across devices

• Manage your credits, purchases, and subscriptions

2. What Happens During a Coaching Session

When you record a session, the system analyzes your speech and presentation in several stages. Your audio and video are processed temporarily during analysis and are not stored permanently.

Audio Processing (Speech Transcription)

Your spoken audio is transcribed using Deepgram Nova-3.

This converts speech into text and extracts speaking metrics including:

• Words per minute (speaking pace)

• Pause count and duration

• Filler word detection

These acoustic signals help generate coaching insights.

Audio is processed either:

• Live (real-time streaming) while you speak, or

• Batch mode after recording.

Body Language Analysis

Your camera feed is used to analyze body language and non-verbal communication.

During recording:

• Frames are captured approximately every 3 seconds.

• Frames are temporarily stored in memory in your browser.

• Frames are compressed to JPEG (~70% quality) with a maximum width of 640px.

• The system maintains a buffer of up to 40 frames using a sampling method that keeps frames representative of the full recording.

For visual analysis:

• Up to 10 frames are analyzed for recordings under 3 minutes.

• Up to 40 frames are analyzed for recordings longer than 3 minutes.

Frames are sent to Google Gemini 2.5 Flash (Vision) for body language analysis.

Raw video recordings and full video streams are not stored.

Coaching Feedback Generation

Your transcript, speech metrics, and body language observations are combined by Anthropic Claude Sonnet 4.6, which generates:

• Performance scores

• Coaching feedback

• A personalized coaching script

• Practice tips

Feedback is streamed to the interface as it is generated.

The system automatically detects the language spoken and generates feedback in the same language.

Optional Voice Playback

If you choose to listen to the coaching script, it may be converted to audio using OpenAI text-to-speech (tts-1, "echo" voice). This feature is optional.

2b. What Happens During an Interview Room Session

The Interview Room is a voice-based mock interview feature. When you use the Interview Room, the system collects and processes data at several stages.

Job Context Collection

Before the interview begins, you provide:

• Your job title (the role you are interviewing for)

• The company name

• A job description (either pasted text or a brief description)

This job context is sent to Anthropic Haiku to generate a structured role analysis that tailors the interview questions to your specific role. The preprocessed context is stored with your session.

Real-Time Voice Interview

During the interview, you speak with an AI interviewer agent named Alex. The voice conversation involves:

• Your voice is transcribed in real time using Deepgram live speech-to-text.

• The AI interviewer's responses are generated by Anthropic Haiku.

• The AI interviewer's voice is produced by OpenAI text-to-speech (tts-1).

The full conversation is transcribed and stored as part of your interview session record. Raw audio from the voice conversation is not permanently stored.

Post-Interview Debrief

After the interview ends, the full transcript is analyzed by Anthropic Haiku to produce a debrief assessment, which includes:

• Performance scores (clarity, presence, specificity)

• A narrative assessment of your interview performance

• A question-by-question log with answer summaries

• A focus area and practice drill for improvement

• STAR method adherence analysis

• Skill coverage and red flag assessments

3. What We Store About Your Coaching Sessions

After a session completes, we store:

• The transcript of your speech

• AI-generated scores and summaries

• Coaching feedback and practice tips

• Speech analysis metrics (pace, pauses, fillers)

• Body language observations

• Session topic, audience, and goal inputs

• Session duration and performance metrics

• Date and time of the session

Interview Sessions (Interview Room)

After an interview session completes, we additionally store:

• The job title, company name, and job description you provided

• The preprocessed role context (role summary, core skills, behavioral themes, challenges, red flags)

• The full interview transcript (both your answers and the interviewer's questions)

• Debrief scores (clarity, presence, specificity) and overall score

• Narrative feedback and practice recommendations

• Question-by-question log with answer summaries

• STAR adherence analysis, skill coverage, theme coverage, and red flag assessments

• Interview duration, question count, and session timestamps

What We Do Not Permanently Store

We do not permanently store:

• Raw audio recordings (from either coaching or interview sessions)

• Raw video recordings

• Full video frames captured during analysis

• Real-time voice audio from interview conversations

Audio, video, and frame data are processed temporarily and then discarded.

4. Payments

Payments are handled securely by Stripe.

While your account is active, we store:

• Your email address

• Stripe customer ID

• Stripe subscription ID (if applicable)

• Credit balance and tier

We never see or store your card details. Stripe handles all payment processing and card information.

If you delete your account, your subscriber profile — including your email, Stripe customer ID, and subscription ID — is permanently removed from our database. Stripe retains its own independent payment records as required by their terms and applicable law; to request deletion of those records, contact Stripe directly.

5. AI Services We Use

To provide coaching analysis, we use trusted third-party AI services:

We do not attach account identifiers (your name, email address, or account ID) as metadata when sending data to these providers. Note, however, that transcripts and video frames are content you produce, and may themselves contain personal information that you choose to speak about or display during a session. That content is processed by the providers solely to generate the requested analysis.

The information shared may include:

• Speech transcripts

• Audio-derived metrics

• Selected video frame images

• Topic and audience context

• Job titles, company names, and job descriptions (for interview sessions)

• Interview conversation transcripts (for debrief generation)

These providers process the data only to generate the requested analysis.

We do not use your sessions to train our own AI models.

5b. How AI Providers Handle Your Data

In addition to our own commitment not to train models on your data, the external AI providers we use have their own no-training policies for the way we access them:

• Anthropic (Claude): Anthropic does not use API inputs or outputs to train its models by default.

• OpenAI: Since March 2023, OpenAI does not use data sent through its API to train its models by default.

• Google Gemini: We access Gemini through Replit's paid proxy, which falls under Google's paid tier. Google does not use prompts or responses from paid services to train its models.

These commitments concern model training only. Providers may still process your data transiently to deliver the requested service and may retain limited operational logs as described in their own policies. This no-training guarantee is independent from the AI processing consent toggle on your Profile page: the toggle controls whether Stacked sends your data to AI providers at all, while this section describes what those providers do (and do not do) with the data once it reaches them.

6. Analytics

We use PostHog Analytics to understand overall app usage.

PostHog is privacy-focused:

• No advertising cookies

• No cross-site tracking

• No personal profiling

Session cookies used for login are functional only.

6b. Sub-processors

We rely on the following sub-processors to operate the service. Each one acts on our instructions and only for the purposes listed below.

6c. Legal Bases for Processing (GDPR Art. 6)

For users in the European Economic Area, we rely on the following legal bases:

Where we rely on consent, you may withdraw it at any time without affecting the lawfulness of processing carried out beforehand.

7. Session Duration Rules

Sessions are categorized by recording length:

Free users are limited to 5 minutes per session.

Paid users can record sessions up to 20 minutes.

8. How Long We Keep Data

• Coaching session results are stored until you delete your account.

• Login sessions expire after 7 days.

• Internal cost and revenue logs are retained for accounting purposes but are anonymized — they contain no name or email address.

When you delete your account, the following are permanently deleted: your user profile, all coaching sessions, your subscriber record (including Stripe IDs and credits), and your active login sessions.

What remains after deletion are anonymized internal records (cost logs and revenue totals) with all identifying fields removed. These are retained solely for financial reporting and contain no information that can be linked back to you.

9. Your Rights

Depending on where you live, you may have the right to:

• Access your data

• Correct your data

• Delete your data

• Export your data

• Withdraw consent

To delete your account, use the Delete Account option in the user menu on the home screen. This immediately and permanently removes your profile, coaching history, and subscriber record from our systems.

To export your data, use the Download my data option on your profile page. This delivers a JSON file containing your profile, subscriber record, coaching history, and interview sessions — everything personal that we store about you.

For all other requests (access, correction, or withdrawal of consent), contact us at:

info@stacked.coach

We aim to respond within 30 days.

10. Children's Privacy

Stacked is not intended for children under:

• 13 years old globally

• 16 years old within the European Union

If we learn that we have collected personal data from a child without appropriate consent, we will delete it.

11. Security

We take security seriously and use industry-standard safeguards including:

• Encrypted HTTPS connections (encryption in transit)

• Encryption at rest for stored data, provided by our managed database and session-storage hosts

• Secure httpOnly session cookies

• Server-side session storage

• Payment processing through Stripe

• Trusted cloud infrastructure

No system can guarantee absolute security, but we work continuously to protect your information.

12. International Data Processing

Our service relies on global infrastructure and AI providers.

Your data may be processed in countries outside your own, including the United States. We take steps to ensure appropriate safeguards are in place when data is transferred internationally.

13. Changes to This Policy

If we make significant changes to this Privacy Policy, we will update this page and revise the "Last updated" date.

14. Contact Us

Questions or concerns?

Email us at:

info@stacked.coach