Frequently asked questions
Honest answers about privacy, what the analysis can and cannot see, and what it costs. If something is missing, write to hello@readbeneath.com.
Last updated: July 15, 2026
Privacy & data handling
Is my conversation private?
Yes. Personal identifiers — emails, phone numbers, addresses, card numbers — are masked before any AI reads a message, and everything moves over encrypted connections. Your conversations are never sold, never used for ads, and never used to train third-party models.
Can I delete my data?
At any time. The raw files you upload are removed from temporary storage once processing finishes, and one click in Settings permanently erases a conversation — or your entire account, including every report built from it.
Do you run ads or share data with advertisers?
No. There are no ad pixels on the site, no data licensing deals, and no third-party trackers watching what you upload. The business runs on subscriptions, not on your data.
What the analysis can and cannot see
Can ReadBeneath tell me who was right?
No — and it will not pretend to. ReadBeneath surfaces observations, not verdicts. Every finding is tied to specific cited messages and paired with a fair alternative reading, because the same exchange almost always supports more than one honest interpretation.
Why did my report say the sample was too thin?
Because honesty beats a confident guess. ReadBeneath enforces hard sample-size floors: manipulation-pattern analysis is skipped below 100 text messages or 7 days of history, named pattern labels are demoted to plain descriptions below 300 messages or 14 days, and relationship-level verdicts are disabled entirely below 500 messages or 30 days. When a question needs more evidence than you uploaded, the report says so instead of fabricating an answer.
How accurate are the findings?
Every finding must cite the specific messages that support it, confidence is capped at 95% — certainty is never claimed — and each observation ships with a charitable alternative explanation. If the evidence is too thin for a pattern, that pattern simply does not appear.
Is this a substitute for therapy?
No. ReadBeneath is a pattern-recognition tool for your own conversations, not a licensed therapist, and it never applies formal mental-health labels to anyone. If a conversation involves your safety or wellbeing, a licensed professional is the right next step.
How it works
What file formats can I upload?
WhatsApp exports (.txt, or the .zip WhatsApp produces), plain-text conversations with per-message timestamps and sender names, CSV files, JSON exports (including the message formats Telegram and Instagram produce), and screenshots of chats, which are read with on-screen text recognition.
How do I export my chat?
Each app has its own path. The step-by-step guides cover WhatsApp on iPhone, Android, and desktop, plus the honest options for iMessage — see the guides at readbeneath.com/guides, or the in-product helper on the upload screen.
What are the analysis lenses?
You pick one focus per analysis: a romantic relationship, personality and communication style, a family or parent-child dynamic, or a just-for-fun read. The focus shapes which specialist agents examine the conversation and how the report is framed.
Pricing
How much does ReadBeneath cost?
The first read is free: a descriptive analysis with pattern observations and two follow-up questions, no card required. Pro is $20 per month and adds Deep Analysis with the full agent council, unlimited follow-up Q&A, per-person conflict breakdowns, and the Chat Recommender.
Can I cancel Pro anytime?
Yes. Pro is a monthly subscription handled through Stripe — cancel from Settings whenever you like and you keep access through the end of the billing period. There are no lock-ins and no cancellation fees.
Ready to see your own patterns?
Upload a conversation and get a free descriptive read in under a minute — cited evidence, a fair alternative on every finding, and a straight answer when the sample is too thin.