A learning management system stores a video of everything you need to know. Napster Learn offers something else entirely

A Digital Twin uses the likeness of a real person (their face, their voice, and their expertise) to teach and guide learners through every tool in the platform. Once published, that Twin teaches a Course, the structured lessons a Twin (or Mentor) teaches, in the form of a live conversation. Learners can interrupt, ask a follow-up, and get answered in real time. 

It sits in Crews, a standing group of experts available on demand. It runs Pulse check-ins, meaning it reaches out to learners on its own instead of waiting to hear back from someone who’s gone quiet. It can also take on different roles to ensure students actually understand the content. For example, Simulate Practice mode can assume the role of the skeptical conversation partner to help challenge students on their knowledge. Then, the Twin can take on the part of the examiner in Assessment mode, testing what someone knows via conversation. 

Even your most qualified human expert only has so much time in the day to answer questions. This is the reason expertise stays trapped with the person who has it and the reason most learners only get information from a recording of someone who cannot hear their question. A Digital Twin closes that distance. It delivers the same expertise, held in as many conversations as there are learners, which is the kind of ratio that creates effective results. 

The value of this product is that it can be in as many places at once, whereas a real-life expert/professor cannot. But that same side of the coin is also the challenge we have with Digital Twins: Everything Napster Learn does happens in the space where the expert is not present to see it; they can’t be standing over the Twin’s shoulder to monitor it. This is why a Twin needs something closer to a standing, ongoing agreement between it and the expert. In other words, it needs a way in, a way to steer, and a way out.

Consent follows the digital twin everywhere it goes.

One property of the Digital Twin setup decides everything downstream: The knowledge uploaded to a Twin follows it everywhere the Twin is used. That is a feature of this product. It is why a single setup session produces an expert who shows up across the learning loop instead of a video that shows up once. It is also why consent to a Digital Twin is not consent to a piece of content. It is consent to a presence.

Organizations that treat each deployment as its own approval — sign off on the Course, sign off on the Pulse cadence, sign off on the Simulate scenario — end up approving nothing meaningfully, because the Twin was the decision. Govern the Twin. The deployments follow.

There is a second thing worth naming early because it resolves half the hard cases before they start: Not every expert in Napster Learn has to be a Digital Twin. A Mentor is a fabricated person, usable anywhere a Twin is usable, with no real likeness attached. When the question is complicated, the Mentor is the answer. Nobody has to be twinned for the platform to work.

How does AI Digital Twin consent work? The five-step setup

To create a Twin, an Admin at the organization clicks Invite on the Digital Twin home screen and enters an email address. The expert receives a link and clicks through to the setup wizard, with no account creation required. They never have to become a Learn user. And then the expert does five things:

Step one: They take a selfie. 

Step two: They sit for an interview, during which the system asks questions to learn about them and their expertise. 

Step three: They edit their own profile overview. 

Step four: They upload the documents that define what their Twin knows. 

‍Step five: They record their consent and then press publish.

‍

A Digital Twin can’t exist without full consent of the user.

A Twin does not exist until the person it represents has completed all five steps. There is no path in which an organization manufactures a Digital Twin of someone who never participated. Consent for this tool is not a box someone checks. The consent here is the process. In other words, when a person takes part in the entire setup of the Twin, from start to finish, they are giving their consent for the Twin to exist. 

There’s an important reason why the process requires the expert to personally do all five steps, though it may seem like it takes a lot of time and energy. In practice, this guardrail protects the buyer as much as the expert. A Twin built around someone who never sat down is not a multifaceted product. It has no voice and leaves open a legal risk for your company because there’s no proof that the twinned person ever agreed to being copied. On top of that, sending the invite and having the real person go through setup is actually the fastest way to do this anyway, so skipping it doesn't even save time.

There’s one part of this that is crucial to know: The invite email itself is the real proof of consent, not the setup wizard. The wizard just proves the person showed up and went through the steps. The email is where the actual agreement needs to be spelled out. That email should clearly state what the Twin will be used for, who will interact with it, how long it'll stay active, and whom to contact if the person wants to change or end any of that later.

“Published” or not is the only switch that matters.

Setup is the beginning of consent, not the end of it. An expert's relationship to their Twin continues for as long as the Twin does, and there are three places that relationship lives.

Knowledge. The PDFs and slide decks uploaded during setup are the substance of the Twin's expertise. Changing what a Twin knows means changing what it has been given. This is the dial that matters most and gets watched least.

Profile. The overview of the person, aka  how they are described to every learner who meets them.

Placement. Where the Twin actually appears. Which Crews include it, and Crews are either draft or published, with only published Crews visible and usable. Which Courses it teaches. Which Pulse schedules it runs. Which Simulate scenarios it plays.

The fact that really matters here is that the “publish” state is the real control surface. It’s as simple as this: A Twin is published or it is not. A Crew sits in draft or it sits in published, and only published Crews are visible and usable. Those look like workflow states on a project board. They are also the line between a person's likeness being in front of learners or not.

How does someone evoke an AI Twin? The three-step process

The test of any consent system is whether “no” still works a year later. Most organizations discover they never defined it, because revocation sounds like one action. In fact, it’s actually three.

Step one: Stop the Twin from appearing. Unpublish it. The Twin goes dark everywhere at once. It will no longer be seen in Crews. It will be taken off its Courses, out of Pulse schedules, and Simulate scenarios. All of this happens without anything being destroyed. 

Step two: Remove the likeness. Delete the Twin. The face and the voice come out of the system, not just out of circulation. This is the request an expert makes when they mean it, and it is a different ask from the first one, which is exactly why both actions exist.

Step three: Decide what happens to the record. Learners completed Courses this Twin taught. Their Talent Profiles hold that history. Does revoking a Twin rewrite the learning record, or does the record stand while the Twin goes dark?

A good default policy: Keep the record of what the user learned, but remove the expert’s face and voice. The learner doesn't lose credit for what they completed. The expert doesn't have their likeness lingering in the system after they've asked for it to be taken down.

And plan for the events that trigger all of this. For example: The subject matter expert resigns, the faculty member takes a position elsewhere, or the person simply changes their mind.

How do you manage Digital Twins in higher education and across companies? And how do you license Digital Twins?

Higher education. A professor’s likeness belongs to them alone, and it is not the institution's to quietly edit. Consider a university offering a Crew of experts to help freshmen navigate their first year. It’s a genuinely good use of the platform and one that outlives any single term. Does a Twin persist after the professor leaves? If someone changes what a Twin knows or teaches, who signs off on that decision? When it comes to adjunct or guest lecturers with an end date, should the Twin also have one? The cleanest answer is to set the faculty member’s Twin’s term length to match theirs. 

Enterprise. The goal here is to turn top performers into Twins so their knowledge reaches everyone. The question that follows is what happens when the top performer resigns from the company or institution.

The answer is to separate the two things employment agreements already separate. The expertise and the documents uploaded is often company IP. The likeness, though, is the experts’ – permanently. There’s an easy solution here: Retire the departing expert's Twin, and create a Mentor on the same knowledge documents. The capability stays. The face goes home with its owner.

Worth saying out loud during setup, too: a Twin used in Simulate Practice mode will be argumentative, meaning it will be challenged in conversations by hundreds of salespeople when they rehearse a hard interaction.. Experts should know that their Twin will be argued with before they publish, not after. 

Selling access to expertise. This is where every question above acquires a price tag. When an expert licenses their Twin rather than lending it, consent becomes a commercial term: scope of use, exclusivity, duration, renewal, and a royalty on Courses generated from their knowledge. And then the term everyone skips: exit. Can a seller withdraw their Twin, and what happens to the organizations mid-engagement when they do?

That market for licensing AI is not hypothetical; it’s coming. And it will belong to whomever figures out permission, ongoing management, and the ability to undo it. The winning company also does that reliably at a large scale and before anyone else.

Six questions to ask any platform before your first Twin: 

1. Who in our organization is allowed to issue a Twin invite?
2. What does that invite promise about scope, audience, and term?
3. After publish, who can change a Twin’s knowledge? And can the person it represents see where their Twin is deployed?
4. What is our default term, and what triggers renewal?
5. How fast do we act when someone asks for their Twin to come down?
6. On revocation, what persists: the Twin, the knowledge, or the learning record?

None of these are hard questions. They are only hard to answer retroactively.

Bottom line: Here’s the real reason to use Digital Twins. 

Your learning management system (LMS) is doing its job. It stores content and tracks compliance. Napster Learn solves a different problem: The conversation layer your LMS was never built to have. This layer is what sets Digital Twins apart. 

A stored document does not represent anyone. A Digital Twin does. It shows up as a person, is treated as a person, and is remembered as one.

Your best experts will lend their likeness to a system that lets them steer it… and take it back. The consent architecture is not the thing standing between you and a Digital Twin. It is the thing that makes anyone willing to become one.

Main takeaways

• A Twin’s value and its risk are the same thing: It can be in as many conversations as there are learners, while the real expert can only be in one. This is why there needs to be an ongoing agreement between the LMS and the expert. 
• Consent is built into the five-step setup process itself. Only the real person can complete it, so a Twin cannot exist without their participation.
• Stopping a Twin from appearing, deleting its likeness for good, and deciding what happens to the learning records it already created.
• Whoever solves consent, control, and revocation first will own this market, not whoever scales fastest.

‍

‍