The New E-learning Production Chain
When AI Tools Meet Authoring Tools
There is a common misconception in organizations that are modernizing their training: the idea that adopting artificial intelligence in e-learning is a matter of choosing the right tool. Buying a voice generation subscription, an avatar platform, an image generator — and that's it, the training is modernized.
The reality is different. The AI tools available today are extraordinarily capable individually. But the real value is not in any isolated tool — it lies in the architecture that connects them to each other and to the authoring tools and LMS platforms that already constitute the core of e-learning production. It is the difference between having a set of excellent musical instruments and having an orchestra that knows how to play together.
This article maps the current state of this new production chain — the categories of tools, what each one solves, and how they relate to the classic work of instructional design. The goal is not to provide an implementation manual (that is the part that requires experience, method, and adaptation to the context of each organization), but to give decision-makers a clear view of the landscape and the decision points that determine the success or failure of a modernization project.
---
The problem AI solves — and what it still doesn't solve
Before mapping tools, it is important to understand where AI adds real value in e-learning production, and where it doesn't.
The production of a quality e-learning module involves a sequence of stages: training needs analysis, definition of learning objectives, content structuring (instructional design), scripting, production of visual and audiovisual elements (image, video, voice), assembly and interactivity in the authoring tool, and publication and traceability in the LMS.
Generative AI has radically transformed the speed and cost of several of these stages — particularly scripting, image and video production, and narration. Tasks that previously required weeks, studio budgets, and specialized teams can today be executed in hours, by a small team, at a fraction of the cost.
But there is one stage where AI still does not replace human judgment: instructional design . The decision about what to teach, in what order, with what pedagogical strategy, with what type of assessment, and adapted to which audience — that decision still depends on human competence, domain knowledge, and experience. AI accelerates execution; it does not replace conception.
This distinction is the foundation of every successful modernization strategy. Organizations that treat AI as a substitute for instructional design quickly produce large quantities of mediocre content. Organizations that treat AI as an accelerator for executing good instructional design produce quality training at unprecedented speed and cost. The difference between the two approaches is, to a large extent, what distinguishes a modernization project that works from one that fails.
---
The categories of the new production chain
The landscape of AI tools applicable to e-learning is organized into distinct functional categories. Knowing these categories — and the reference tools in each one — is the first step for any decision-maker evaluating the modernization of their training production.
Scripting and content structuring
Large Language Models (LLMs) are today capable of substantially assisting in the transformation of raw technical content — manuals, regulations, procedural documentation — into script structures suitable for e-learning. They can suggest module structures, write first narration drafts, generate scenario variations, and propose assessment questions aligned with the content.
The value lies in eliminating the blank page and accelerating the first draft. The limitation is that the raw output of an LLM is rarely suitable for direct use — it requires curation, alignment with specific pedagogical objectives, adaptation to the organization's tone and context, and rigorous factual checking, especially in regulated sectors where precision is critical.
Image production
AI image generators allow for the creation of illustrations, scenarios, characters, and customized visual elements without resorting to image banks or illustrators. For sectors with specific visual contexts — industrial environments, naval equipment, clinical contexts — where generic image banks rarely have what is necessary, this capability is particularly valuable.
The integration of these generators into production workflows, however, raises questions of visual consistency (maintaining a coherent style throughout a course), contextual suitability, and usage rights — issues that determine whether AI-generated visual production adds quality or introduces incoherence.
Video production and editing
This is probably the category where AI has had the most visible impact. Avatar tools (virtual presenters), animation, video translation with lip-syncing, and transcription-based editing have completely transformed the economics of training video production. A video that previously required a studio, actor, and production team can today be produced digitally, updated by editing text, and localized into multiple languages without re-recording.
The choice between the various approaches — realistic avatar vs. animation, automatic translation vs. native multilingual production — depends on factors that go far beyond the features of each tool: the type of content, the audience, the cultural context, update requirements, and the organization's localization strategy.
Voice-over and narration
AI voice synthesis has reached a quality level that makes it viable to replace human narration in most e-learning contexts. Voice cloning allows for maintaining vocal consistency across extensive catalogs, and support for European Portuguese has significantly improved. Narration production has ceased to be a cost and time bottleneck.
The relevant decisions in this category are less obvious than they seem: when to use synthetic vs. human voice, how to manage voice cloning in compliance with legal and ethical requirements, and how to integrate narration into the production flow without creating dependencies that make future updates difficult.
Assembly, interactivity, and publication
At the end of the chain are the classic authoring tools — Articulate, iSpring, Lectora, or open-source solutions — where all the produced elements combine into interactive, traceable learning experiences publishable on the LMS. This layer was not replaced by AI; it was fueled by it. Authoring tools remain the convergence point where instructional design materializes.
---
The real challenge: integration, not tools
Here we reach the central point of this article — and the reason why adopting AI in e-learning is more complex than buying subscriptions.
Each of the categories above contains multiple tools, each with distinct strengths, limitations, pricing models, and suitability profiles. Choosing the right tool in each category is already a challenge. But the real challenge — the one that determines whether a modernization project produces results or wastes resources — is the integration of all these pieces into a coherent, efficient, and sustainable production chain.
This integration raises questions that have no single answer, and whose correct resolution deeply depends on the specific context of each organization:
How to design the workflow so that the output of one tool efficiently feeds the next, without constant manual rework and without loss of quality in the transitions between stages.
How to maintain consistency — visual, voice, tone, brand — throughout a catalog produced with multiple AI tools that, by default, do not "talk" to each other.
How to guarantee pedagogical quality when production speed increases dramatically — avoiding the real risk of producing large volumes of formally impressive but pedagogically weak content.
How to manage costs strategically, when each tool has its own pricing model (credits, subscription, per user) and uncoordinated combination can result in total costs much higher than necessary.
How to ensure compliance — GDPR, usage rights for cloned voice and image, sectoral compliance requirements — in a production chain involving multiple AI providers with distinct data policies.
How to prepare for sustainability — choosing a production architecture that allows for easy content updates, scalability when necessary, and avoiding dependencies that make the organization hostage to a specific tool or vendor.
How to adapt everything to the sector — because the ideal production chain for banking compliance training is different from the ideal for maritime training, which is different from the ideal for corporate onboarding. There is no universal architecture; there is the right architecture for each context.
---
Why this is a consulting job, not an acquisition job
The natural temptation for many organizations is to treat e-learning modernization as an acquisition problem: identify the tools, buy them, and train the team to use them. This approach produces, at best, partial results — and, at worst, a significant investment in underutilized tools and inefficient workflows.
The reason is simple: the value is not in individual tools, which are available to any organization that wants to acquire them. The value lies in the knowledge of how to combine them — a competence built with practical production experience, in-depth knowledge of the strengths and limitations of each tool, mastery of instructional design, and understanding of the specific requirements of each sector.
It is a competence that combines three types of knowledge that rarely coexist in a single internal team: pedagogical knowledge of instructional design, technical knowledge of the tools and their integration, and sectoral knowledge of specific compliance and context requirements. The scarcity of this combination is precisely what makes specialized consulting valuable — not as a substitute for internal capacity, but as an accelerator of its construction and as a guarantee that structural decisions are made correctly from the start.
A well-conducted modernization project does not just deliver an assembled production chain. It delivers an empowered organization — with the right architecture for its context, with a team trained to operate it, and with the autonomy to evolve its training production as tools and needs change.
---
The Shift2new approach
Shift2new works with organizations in regulated sectors — banking, insurance, healthcare, maritime training — where instructional design quality and compliance rigor are non-negotiable, and where production efficiency is increasingly critical.
Our approach starts from a premise: technology is the means, not the end. Before any tool recommendation, we seek to understand the organization's training objectives, its regulatory context, its internal capacity, and its long-term strategy. From there, we design the appropriate production architecture — the specific combination of authoring tools, AI tools, and LMS platforms that serves the organization's goals, integrated into an efficient and sustainable workflow.
We don't sell tools. We design training production systems, empower the teams that operate them, and accompany their evolution. The knowledge of how everything comes together — the part that makes the difference between an investment that works and one that is wasted — is precisely what we offer.
For organizations considering modernizing their e-learning production with AI, and wanting to do so with method rather than by trial and error, the starting point is a conversation about your specific context.
---
Conclusion
The new e-learning production chain — where AI tools for scripting, image, video, and voice integrate with authoring tools and LMS platforms — represents the most significant transformation in digital training production since the emergence of e-learning itself. Organizations that master this chain will produce better, faster, and cheaper training than ever before possible.
But mastering the chain is not a matter of acquiring tools. It is a matter of architecture, method, and integrated knowledge — of knowing not just what each tool does, but how they all come together to serve concrete pedagogical objectives in a specific context. That knowledge is the real asset. And it is precisely where the value lies.