From AI experiments to professional video pipelines – with Akmalidin Alimov.

Creating AI-generated video has become easy, but making it actually look professional is still genuinely difficult. In this week’s episode of Vad fan ska man ha AI till?, I sit down with Akmalidin Alimov, a former telecom engineer at Ericsson who now works full-time with commercial AI content and automated creative workflows. We discuss what it actually takes to deliver professional results and why a single prompt is rarely enough. It is about building complex pipelines where ideas, images, video, music, and different AI models work seamlessly together, and why he sometimes needs 27 iterations on a single clip before it actually feels human.We break down the tools required to build scalable workflows in practice, exploring how platforms like Higgsfield, Magnific, Claude, and ChatGPT serve fundamentally different purposes. However, the conversation extends far beyond content production. Akmalidin’s company is a prime example of building an AI-native organization: they coded their own course platform using Claude Code and created an autonomous agent to manage their Meta Ads instead of outsourcing the work. This episode highlights the fundamental difference between merely playing with generative AI and actually building workflows that function in reality as repetitive specialist work becomes automated.Perhaps the most important takeaway from our conversation is that the technology itself does not solve everything. Context matters far more than the specific model you use, and professional production still heavily relies on human creativity. AI does not automatically know what feels natural or human. We still have to teach it.Akmalidin Linkedin:https://www.linkedin.com/in/akmalidinalimov/