An executive's calendar is a battlefield. Board meetings, client calls, crises that arrive unannounced. Asking them to commit to a fixed two-hour learning slot is asking them to predict the unpredictable.
I taught in universities for years before I understood this. University students can organise their lives around a class schedule. Executives cannot. When the pandemic forced me online, I expected the format to be a limitation. Instead, it removed one: the flexibility to learn at odd hours, in short bursts, turned out to be exactly what busy professionals needed.
I taught in universities for years before I understood this. University students can organise their lives around a class schedule. Executives cannot. When the pandemic forced me online, I expected the format to be a limitation. Instead, it removed one: the flexibility to learn at odd hours, in short bursts, turned out to be exactly what busy professionals needed.
The other side of the problem
University teachers are busy too. Admin and bureaucracy are relentless. Updating a syllabus can take years to work through approval committees. By the time new content is actually delivered, it is often already outdated. If you update your materials for the latest developments this year but can only teach them two or three years later, you are permanently behind. It is frustrating, and it shows in the quality of what students receive.
At VARi, we are not bound by those constraints. Our courses on AI are updated weekly, sometimes more often, as soon as we learn and test a new development relevant to our students. We stay grounded in the foundational concepts that do not change: what LLMs actually are, the mathematics behind generative and non-generative AI. But the practical elements, the use cases, best practices, and available tools, we update as soon as we have verified their quality and usefulness.
In live courses, making these changes is straightforward. The real shift comes with our asynchronous, self-paced course. And that shift would not be possible without AI.
Because we do not record in the traditional sense. We use avatars.
At VARi, we are not bound by those constraints. Our courses on AI are updated weekly, sometimes more often, as soon as we learn and test a new development relevant to our students. We stay grounded in the foundational concepts that do not change: what LLMs actually are, the mathematics behind generative and non-generative AI. But the practical elements, the use cases, best practices, and available tools, we update as soon as we have verified their quality and usefulness.
In live courses, making these changes is straightforward. The real shift comes with our asynchronous, self-paced course. And that shift would not be possible without AI.
Because we do not record in the traditional sense. We use avatars.
The recording problem
If you have ever produced educational materials, you know how much time disappears into logistics. Setting up lights, microphones, and cameras. Countless retakes because the instructor made a mistake, sneezed at the wrong moment, or forgot to clip on the lapel mic. Then comes post-production. And all of this assumes everyone involved can find a mutually convenient time.
A few minutes of genuinely high-quality tutorial, with proper structure, clear narration, and educationally useful visuals, can take hours or even days to produce end to end. I am not talking about the informal, screen-recorded style common on YouTube. I mean something you would be proud to put in front of a paying professional.
In a field moving as fast as AI, by the time you finish recording, the content may already be out of date. So you face a choice: go through the whole process again, risking the same outcome, or release something you know is not as good as it should be.
We do not like that compromise.
A few minutes of genuinely high-quality tutorial, with proper structure, clear narration, and educationally useful visuals, can take hours or even days to produce end to end. I am not talking about the informal, screen-recorded style common on YouTube. I mean something you would be proud to put in front of a paying professional.
In a field moving as fast as AI, by the time you finish recording, the content may already be out of date. So you face a choice: go through the whole process again, risking the same outcome, or release something you know is not as good as it should be.
We do not like that compromise.
A new opportunity
Using high-quality avatars, we can update a video in minutes. No scheduling. No setup. That frees us to spend our time where it matters most: on the quality of the content itself. What to teach. How to teach it. How to structure it for maximum efficiency.
Instead of wondering whether my hair looks acceptable in this take, I can focus on creating a clear explanation and modify it as many times as needed. Each iteration takes minutes, not hours.
And the quality? Avatars used to look tacky. Some still do. But with the right tools and enough care, they are good enough, and certainly better than plenty of "live" instructors I have seen in educational recordings from even top institutions.
My personal quality test was my own family. They did not realise it was an avatar rather than a real recording of me. If my own mother cannot tell the difference, that is good enough for me.
Instead of wondering whether my hair looks acceptable in this take, I can focus on creating a clear explanation and modify it as many times as needed. Each iteration takes minutes, not hours.
And the quality? Avatars used to look tacky. Some still do. But with the right tools and enough care, they are good enough, and certainly better than plenty of "live" instructors I have seen in educational recordings from even top institutions.
My personal quality test was my own family. They did not realise it was an avatar rather than a real recording of me. If my own mother cannot tell the difference, that is good enough for me.
An honest assessment
Avatars are not perfect. Neither are traditional recordings. The difference is that avatars will only get better, and they already solve the update problem that makes so much online education stale within months of release.
There is also more to an educational video than a talking head. Diagrams, visualisations, worked examples: these are usually more important for what a learner actually takes away. The presenter is just one element, and often not the most critical one.
So yes, we are using avatars in our self-paced course. If you are not sure how you feel about that, watch our free taster session and judge for yourself.
There is also more to an educational video than a talking head. Diagrams, visualisations, worked examples: these are usually more important for what a learner actually takes away. The presenter is just one element, and often not the most critical one.
So yes, we are using avatars in our self-paced course. If you are not sure how you feel about that, watch our free taster session and judge for yourself.
Created with
