Five minutes of you on camera.
In return: an AI version of you that can render an unlimited number of branded short-form videos with your voice, your face, your mannerisms, and your brand kit. Forever.
That is Twin Mode. This is the complete picture of what it is, how the 5-minute enrollment works, what the twin can and cannot do, and the ethics we built around it.
What Twin Mode actually is
Twin Mode is a digital twin of you. Trained from a 5-minute enrollment session. Rendered by Hedra Character-3 (primary) or HeyGen Avatar IV (premium). Voice cloned via ElevenLabs.
The twin is not a generic avatar. It is you specifically. Your face. Your voice. Your speaking cadence. Your hand gestures. Your micro-expressions. When you watch a Twin Mode render, the question "is this me" should have an answer of "yes, this is me, this looks like a moment I forgot recording."
The twin is yours. The data lives in your account. The likeness rights stay with you. The training data is encrypted at rest and not used to train any general-purpose model. We documented the full ethics framework separately because it deserves its own treatment.
The 5-minute enrollment
The enrollment session is a guided recording inside the ENCORE app. Mobile or desktop. Single take.
The flow:
Minute 1: lighting and framing. The app shows you a target box. You position your phone or webcam at chest level, 18 to 24 inches away. The app checks the lighting, prompts you to adjust if needed.
Minute 2: consent and disclosure. Two-step confirmation. You confirm you are the person being recorded, you authorize ENCORE to use this data to train your twin, you confirm the twin will carry a watermark per platform compliance rules (TikTok, Meta, YouTube require AI disclosure for any avatar content).
Minutes 3 to 5: the recording. A 3-minute scripted recording. The script is designed to capture your range. 40 seconds of neutral talking. 30 seconds of high-energy delivery. 30 seconds of slow narrative pacing. 20 seconds of laughter and natural reactions. 20 seconds of side-profile angles. 30 seconds of you reading numbers (this trains the voice model on quantitative content). 30 seconds of you reading a sentence in three different emotions.
That is it. 5 minutes of footage. About 800 visual frames per second, so roughly 240,000 frames total. Plus voice data.
The processing takes 4 to 6 hours. You get a notification when the twin is ready. You can preview, approve, or request a re-enrollment if the likeness misses.
What the twin can do
Three categories of use.
Standalone twin renders. You provide a script (text or voice memo). The twin renders the full video with your face, voice, and brand kit. No source footage needed. This is useful when you have an idea but no time to shoot.
Twin overlay on Library footage. You provide your real footage. The system uses the twin to fill in moments you did not capture (a transition line, a callout, a closing CTA). The cut blends real you and twin you. Useful when your shoot was close but not complete.
Pure twin generation. Script in. Twin video out. Cross-channel variants generated automatically. Useful when you need to ship content during a week you cannot record.
Studio tier gets 30 minutes of Twin Mode per month. Agency tier gets 120 minutes plus 5 twins per workspace.
The math on what 30 minutes of Twin Mode looks like in actual content output is here.
What the twin cannot do
Three honest limitations.
It cannot replace a strong on-camera moment. If your top-performing content comes from spontaneous reactions, real emotion, and unscripted moments, the twin will feel flat next to that. The twin is best for evergreen formats. Hook intros. Brand explainers. Recurring segments. CTAs.
It cannot fix bad scripts. The twin renders what you write. If your script is generic, the twin video will be generic. The twin is an execution tool, not a creative tool.
It cannot grow a new audience by itself. The twin keeps your existing audience engaged when you cannot record. It does not magically attract a new audience. Audience growth still comes from idea quality and channel-fit, not from execution speed alone.
These limits are why ENCORE is Library Mode + Twin Mode, not Twin Mode alone. The real footage wins for spontaneity. The twin wins for consistency and volume.
The fidelity question
The single biggest question I get on calls is "how good is the twin actually."
Honest answer in three parts.
Day 1 fidelity: 92 to 95% on most enrollment sessions. The twin is convincingly you to your existing audience. Cold viewers might catch a tell within 30 seconds.
Day 30 fidelity: 96 to 98%. The system has more signal about your micro-expressions because you have rendered more videos. The twin tightens.
Day 90 fidelity: 98 to 99%. Even your friends will struggle to spot the twin. We have published 12-week fidelity tracking data showing this curve.
The detailed week-by-week fidelity data is in the voice and likeness fidelity post.
Why the twin works for "one shoot, 7 days of content"
Here is the practical workflow.
You shoot one day a week. 60 to 90 minutes of raw footage across 4 to 8 short concepts. That is your real-you content for the week.
Then Monday through Sunday, you ship 1 to 2 videos a day. Some are Library Mode cuts from your shoot. Some are Twin Mode renders for the topics you did not capture, or the daily check-in posts that do not need a fresh shoot.
The math: 6 hours of shooting time per week (one day, with prep and travel) produces 30 to 50 channel-specific videos via the Library + Twin combination. Without Twin Mode, the same 6 hours produces about 12 to 15 videos before the math runs out of source material.
The 3 to 4x lift in output per shoot hour is what unlocks the "one shoot, week of content" promise.
The full Sarah K case study walks through this on a real schedule.
The ethics line
We draw a hard line on three things.
One. The twin is yours. We do not use your training data to train any general-purpose model. We do not share it. If you cancel your subscription, you can request deletion within 7 days.
Two. Platform compliance is enforced at render time. Every Twin Mode video carries an AI-disclosure watermark and metadata flag. We do this even when the platform does not require it (LinkedIn, Pinterest currently do not enforce). We do it because audience trust is a long-term asset and the cost of breaking it is permanent.
Three. We do not allow twins of public figures, deceased persons, or anyone who has not personally enrolled. Consent is verified at the recording layer (the device camera, voice match against the enrollment audio). Attempts to enroll someone other than the account owner are rejected.
The full ethics framework with the consent flow is in this post.
Why this is a switching cost
The twin is the deepest part of the ENCORE moat.
If you spend 8 weeks training a twin in our system and then try to leave, you start from scratch. The competitor's enrollment will not match the fidelity, because their system has not learned your micro-expressions across 50 prior renders. You have to start over.
This is intentional, but it is also a natural property of the data architecture. The twin is not a static asset. It is a continuously refining model that gets better with every video you ship.
By month 3, the cost of recreating your twin elsewhere is operationally prohibitive. Most creators in the private beta describe the twin as "the thing I would not switch away from even if you raised the price."
The full switching-cost argument is here.
Who should use Twin Mode
Three creator types where Twin Mode pays back fastest.
Creators with high upload frequency. If you ship 5+ videos a week, the Twin saves hours per week. ROI in week 1.
Creators with brand consistency requirements. If your content style is highly consistent (educational, talking head, formal voice), the twin lands well. The fidelity is highest for consistent-style creators.
Creators with travel or burnout risk. If you regularly miss your posting cadence due to life events, the twin fills the gap with on-brand content. Cadence becomes a system, not a discipline problem.
Who should not use Twin Mode: creators whose entire value prop is spontaneity and live reaction. The twin will undercut the brand if used for that format.
What to do with this
Twin Mode is gated to Studio ($129) and Agency ($349) tiers. The Free and Pro tiers are Library Mode only.
If you are reading this and on the fence about Studio, the twin is the upgrade trigger for most creators in the private beta. 63% of surveyed creators cited Twin Mode specifically as the reason for the upgrade.
See the Studio tier features here or reserve your waitlist spot to get early access.