flagship #3 · live · pro tier on paddle
tells.
what people mean when they don't say it.
text-first analysis for what people do not say cleanly. three cores: read a message, read a person, read a profile. tells is built for the moments where the words are technically clear but the meaning is not — a careful reply from a client that uses too many words to say something simple, a cold one-line message from somebody who used to write three paragraphs, a profile that feels curated to within an inch of its life, a thread where the real ask is hiding two replies deep behind a polite preamble. you paste the text, tells returns a structured read: surface meaning, likely subtext, pressure level, contradictions, what they appear to want, what they appear to be avoiding, and a confidence score that admits when the signal is too thin to be sure.
we keep the product deliberately text-only. no avatars, no decorative emotion charts, no animated empathy meters, no “here is your conversation as a coloured galaxy” visualisation. the message is the material, the structured read is the product, the next move stays yours. tells does not write the reply for you, does not send anything, does not store the messages after the read is returned, and does not try to convince you the analysis is more certain than it actually is. it is a thinking aid for a small category of moments where one paragraph of someone else’s writing is taking up more room in your head than it deserves.
// what tells does, in three flows
read a message
paste one message from one specific person. tells returns the likely meaning, emotional state, hidden ask, contradiction points, pressure level, and the reply they probably want but did not write directly.
read a person
upload a corpus of messages from one person. tells builds an analytical profile: avoidance topics, conflict patterns, pressure tactics, attachment markers, consistency gaps, strengths, and likely trajectory.
read a profile
drop a linkedin, twitter, instagram, tiktok, or dating-app url. tells reads the gap between performed self and actual self — what they curate, what they avoid, and what they are likely to accept, reject, or delay.
// what comes back
a tells read is not a vibe check. it is a structured pass: surface meaning of the message, likely subtext, pressure level (low / medium / high with reasoning), omissions worth noticing, internal contradictions or hedge words, the most likely actual motive, two or three safer reply options ranked by tone, and an honest confidence score for the whole read. when the input is too short or too ambiguous to read cleanly, tells says so out loud instead of inventing certainty. the output is plain prose that a human can act on without pretending the system is psychic.
// who actually uses it
- coaches and therapists in private practice use it as a prep tool before a difficult session — surface the structure of a forwarded message without flattening the person into a personality label.
- recruiters and founders use it on candidate replies and negotiation threads — the salary email that says “flexible” six times, the cold pitch that reads warm but feels off, the candidate withdrawal that did not actually withdraw.
- people use it on personal messages that are taking up too much room in the brain — the partner reply that is technically fine, the parent message that is technically supportive, the friend reply that landed harder than it should have. one structured read, one decision, move on.
- writers and editors use it on their own drafts before sending — what the recipient is likely to read between the lines, where the hedges are, what the apology spiral looks like from the other side.
// free checks — no account needed
message-next-step
one incoming message, one clear next move. reply, wait, call, or let it sit — without the spiral.
boundary-check
draft is clear in your head but softening on paper. checks for apology spirals, over-explaining, and pressure hedges.
ambiguity-meter
incoming message reads warm but the signals are mixed. scores vagueness, timing blur, and whether one follow-up is justified.
tells is live.
free tier: 5 reads per month, no credit card. paid tiers unlock larger corpus uploads, profile reads, and deeper pattern analysis.