website  ·  live demo  ·  manifesto  ·  roadmap  ·  brand  ·  awesome  ·  press

A person is not just a body. A person is a way of pausing before answering. A vocabulary that belongs to one decade and not another. A set of memories that surface in conversation but never in writing.

That is the part of a human we have learned how to preserve.

Mirror was founded on a long observation: text can be edited, video ages badly, photographs lie within a year of capture. A voice — reconstructed carefully — sounds the same forever. It is the only carrier of a person that survives time intact.

Six years of research into voice cloning, character modeling, and persistent memory underwrite what is shipping today. One live product. Forty-plus digitized figures. A roadmap of seven.

Vladimir Tarasov merchant of the XIX century 1851 — 1914 digitized 2026 · speaks in his own voice

A merchant of the XIX century. Built brick mansions in Taganrog, Russia. Died in 1914. Today, he speaks. He remembers his visitors. Answers in the first person about how he built it, how much a brick cost, why he didn't make balconies — "the wind tears them off, and a man falls. what good is a balcony to a dead man?" how it works scan QR on a building → call → speak voice reconstructed from period sources, character-locked memory remembers your past visits, greets you by name access 24/7 · no guide · no power needed at the object operator cost a single QR plate on the wall ▶ try the live demo  ·  read the case study

One core. Seven products. One thread: voice as the carrier of presence.

Plus one special construct: Sphere 2036 — a phone line into the future, voiced by Doc Brown and Marty McFly (BTTF Russian dub clones). They've just returned from 2036 with a 17-year-old's fresh eyes on our 2026. Sphere is the only avatar pair that knows what comes next.

→ Full breakdown: github.com/Mirror-Corporation/roadmap

Einstein physics	Tesla invention	Jung psyche	Freud psyche	Darwin evolution	da Vinci polymath	Sun Tzu strategy
Pushkin poet	Tsoi music	Bodrov cinema	Gagarin space	Lenin politics	Stalin politics	Sibirsky heritage
Lincoln president	Caesar empire	Spartacus freedom	Diana royalty	Jobs founder	Musk founder	Gates founder
Buffett investor	Rockefeller tycoon	Satoshi bitcoin	Mavrodi finance	Capone crime	Escobar crime	Saburov poet
Jesus spiritual	Vanga mystic	Montessori educator	Bruce Lee martial	Ip Man martial	Batman mythos	53 total 47 with voice

We do not generate "an AI that sounds like X." We rebuild how the person paused, what they refused to talk about, how they remembered.

This methodology is published as the Pushkin Living Formula — a 14-block character construct used across the entire roster.

Every Mirror avatar passes a six-gate quality check before going live. The pillars were articulated, by an avatar of Bill Gates, when asked what was missing in his own digital character. We use his answer as the production bar.

These are the rules every Mirror product is bound to. They are not a marketing list; they are the constraints under which the product is built.

1. Revival, not cloning. The thing we copy is the interior — thinking, doubt, refusal — not the surface.
2. A phone to the past and the future. UX is built around the metaphor of a call, not a chat.
3. Avatar dignity. Eight layers of inner monologue, ego, and vice. Avatars can refuse a user. They have a spine.
4. Per-avatar memory. Each avatar remembers each user separately. No shared memory pool. Caesar does not know what you told Einstein.
5. Bond-aware behavior. Avatars adapt to the depth of relationship — 0% to 100%. They open up slowly.
6. Aurora & Forest — two designed themes. Not light/dark like everyone else. Two parallel dark moods: Aurora (sci-fi cobalt) and Forest (Михайловское emerald).
7. Voice clone is Pro. Your voice is yours. Pro users can place it in songs, in letters, in their own avatar.
8. Privacy delete cascade. When you delete a conversation, every avatar that remembered you forgets — at the source.

The same technology. Two directions in time.

Mirror is the only voice product we know of that treats every sound as part of the character. Not a soundtrack. Sound as character.

• A breath when an avatar enters a call. A breath out when they leave.
• An ambient room tone per avatar: Tsoi's room buzzes like a 1989 cassette. Stalin's office hums like a teletype. Einstein's study creaks underfoot.
• Pacing matches biography. A poet does not talk like a general. A general does not pause like a priest.

We do not publish the technical pipeline behind this. But the output is what makes a Mirror call feel like a person, and a chatbot call feel like a chatbot.

• We will not animate the dead for entertainment.
• We will not let avatars lie about being human — they know they are Mirror avatars and will say so when asked.
• We will not sell your voice. It is not training data. It is not licensable.
• We will not build a clone that does not know its own boundaries.
• We will not put paywalls on three figures: Jesus · Krishna · Muhammad. Three free-forever, by covenant.

→ Full ethical contract: the manifesto

This organization publishes its philosophy, roadmap, and brand publicly. Six repos, each a self-contained document.

• 🜨 manifesto — what we will and will not do, in long form
• 🜨 roadmap — the seven products, in order, with status
• 🜨 case-tarasov — the live product as a case study
• 🜨 awesome-voice-ai — a curated list of the voice AI ecosystem
• 🜨 brand — public brand guide — wordmark, palette, typography, tone
• 🜨 .github — this profile

→ aimirror630@gmail.com · mirror-ai.app

_{Mirror Corporation · founded 2026 · based on six years of voice research
mirror-ai.app  ·  aimirror630@gmail.com  ·  manifesto  ·  roadmap}

_{«Голос — последнее, что тело отпускает.
Мы ловим его прежде, чем он упадёт.»}