Q&A: Frontline

I have been reflecting on the Frontline announcement and would appreciate some deeper answers from Articulate about the thinking behind the product.

Much of the initial discussion has focused on whether Frontline represents a threat to instructional designers. That is understandable, but it is not my main concern. My concern is what Frontline communicates to organisations about the value of learning expertise and about what designing effective workplace learning actually involves.

I have worked in adult and digital learning for approximately 30 years. I am not resistant to new technology, nor do I believe that learning professionals should be the only people permitted to create learning resources. Subject-matter experts and operational teams should absolutely be able to share knowledge quickly. AI can also provide valuable support when it is used appropriately.

However, Frontline appears to go considerably further than making an authoring tool easier to use.

Rise lowered the technical barrier to producing online content. It enabled people without programming or advanced development skills to create clean, responsive courses. That inevitably led to more content being produced outside specialist L&D teams, but someone still had to make the learning decisions. The tool did not claim to diagnose the performance problem, determine what people needed to do differently, decide which material was relevant, select an appropriate learning strategy or establish whether training was the right intervention.

Frontline appears to position Nova as capable of taking on at least some of that professional judgement.

The proposition seems to be that someone can upload a presentation, document, procedure or recording, or describe what people “need to know”, and Nova will convert it into a coordinated training kit containing the appropriate combination of guides, quick references, presentations, interactive videos, scenarios and other assets. Articulate describes this as AI-guided training creation “grounded in learning science.” The website also states that users do not need an instructional design background to get started.

I have seen this with a platform called Didask, and its a disaster. Uploading bias content and asking it to create something goes against all learning best practises. Can Frontline determine and say "Actually, no training is required"?

That is where my frustration begins.

The difficult part of workplace learning has never been converting documents into attractive content. The difficult part is establishing whether a learning need exists, what is preventing the required performance, what people must be able to do in the workplace, what practice they require and how we will know whether anything changed.

Source material does not answer those questions.

A subject-matter expert may know the subject extremely well, but that does not necessarily mean the expert can identify the underlying performance need or design an effective learning intervention. An SME can provide accurate and relevant knowledge. A learning professional should determine how, why and whether that knowledge needs to be turned into learning.

If the process begins with “upload your source material,” the risk is that the source material becomes the curriculum. If the process begins with “describe what people need to know,” the risk is that knowledge acquisition becomes the assumed objective. Yet workplace performance frequently depends on recognition, judgement, decision-making, practice, feedback, environmental support and the ability to apply knowledge under realistic conditions.

Frontline may be technically impressive. My concern is that it could encourage organisations to believe that generating learning content is equivalent to designing learning.

I would therefore welcome detailed answers to the following questions.

1. What does Articulate mean by “grounded in learning science”?

Which specific research, principles or established bodies of evidence inform Nova’s decisions?

How does Nova determine which learning format is appropriate for a particular need? How have those decisions been validated, and against what professional or performance criteria?

Has Frontline been independently evaluated to establish whether its outputs produce better retention, transfer or workplace performance than conventional documents, presentations or AI-generated content?

“Learning science” is an extremely broad expression. If it is a central product claim, customers should be able to examine what it means and the evidence supporting it.

2. Does Frontline analyse a performance need, or does it transform supplied content?

Can Nova distinguish between a knowledge gap, a skill gap, a motivation issue, a process problem, poor system design, insufficient resources and an organisational barrier?

Can it conclude that training is not the appropriate solution?

If a manager uploads a 70-slide presentation and asks for training, will Frontline challenge the assumption that the presentation should become training? Will it identify irrelevant content, missing information or unsupported assumptions? Or will it primarily organise and transform whatever it has been given?

This distinction is fundamental. Content transformation and performance analysis are not the same activity.

3. How does Nova establish what learners need to do?

Does Frontline require the creator to identify observable workplace behaviours and performance outcomes, or can a kit be generated from a general description of what learners should “know” or “understand”?

How does it account for learners’ existing knowledge, experience, roles, working environments, language needs and barriers to performance?

If these factors are not examined, how can the system select an appropriate learning strategy rather than simply an appropriate presentation format?

4. How are meaningful scenarios created?

I welcome the inclusion of scenarios and conversational practice because learning should involve more than consuming information. However, a scenario is not effective merely because it contains a conversation and several options.

Good scenario design depends on credible context, realistic ambiguity, meaningful choices, plausible consequences and feedback that helps learners refine their judgement. It often depends on understanding the social, cultural and operational environment in which the decision will be made.

How does Nova establish that context? How does it avoid producing obvious choices, simplistic distractors and generic feedback? How are scenario quality and authenticity evaluated before the content reaches learners?

5. Where exactly does L&D governance sit within Frontline?

Articulate says Frontline allows learning expertise and standards to extend across the organisation. What mechanisms make that happen?

Can L&D establish enforceable design standards, required workflows, approved learning patterns and publishing controls? Can an organisation require professional review before a kit is distributed? Can different levels of creation and publishing permission be assigned?

Brand colours, logos and approved assets are useful, but brand consistency is not learning governance. A resource can be perfectly on-brand while still being educationally weak, unnecessary, inaccurate or inappropriate.

If professional review is optional, Frontline is not necessarily extending L&D expertise. It may simply be allowing people to bypass it.

6. How is effectiveness measured?

Articulate refers to completion, engagement, survey data, questions submitted to the AI Tutor and, in places, business outcomes and ROI.

Completion and engagement show that people accessed or interacted with something. Surveys provide perceptions and self-reported responses. Questions submitted to a tutor may reveal uncertainty. None of those measures, by themselves, demonstrates changed behaviour, improved performance or business impact.

What evidence allows Frontline to connect participation with workplace outcomes? Can it incorporate operational performance data, manager observation or evidence of transfer? If not, how should customers interpret claims about “real ROI” or knowing that the training “worked”?

7. How is the AI Tutor governed?

Is the tutor restricted entirely to approved source material? How does it respond when the source information is incomplete, ambiguous or incorrect?

Can creators examine the answers it has given, identify unsupported responses and correct them? Is there an audit history? Can high-risk questions be escalated to a human expert?

These issues become particularly important in compliance, finance, safety, technical and regulated environments, where a confident but inaccurate answer can have consequences.

8. What happens when source material changes?

The ability to update every asset in a kit from a single instruction is attractive, but it also introduces risk.

Does Frontline show exactly what changed in every affected asset? Is human approval required before those changes become visible to learners? Can an organisation retain previous versions, compare changes and roll back an update?

Automatically propagating an error would be just as efficient as automatically propagating an accurate update.

9. How are accessibility and localisation quality assured?

Does every Frontline format conform to WCAG 2.2 AA by default? What accessibility decisions remain the responsibility of the creator?

When a kit is translated into more than 80 languages, who validates technical terminology, local legal meaning, cultural context, scenario realism and the suitability of AI-generated narration?

Translation at scale is valuable, but speed and linguistic availability should not be presented as equivalent to accurate localisation.

10. How does Articulate see the future role of the learning professional?

This is perhaps my most important question.

Articulate built much of its success through a professional community that used Storyline to move beyond passive page-turning courses. Layers, states, triggers and variables allowed designers to create practice, simulations, branching decisions and meaningful consequences without becoming programmers.

Learning also appears to be one of the few professional fields in which expertise is routinely treated as optional. Non-learning colleagues regularly tell learning professionals what they should create and how they should work. We would not normally tell accountants how to apply accounting standards, yet almost everyone seems to believe that experiencing education qualifies them to direct learning design. Frontline risks reinforcing that belief.

Frontline is now being marketed directly to business teams on the basis that they do not need an instructional design background because the learning science is built into the product.

Does Articulate believe that professional learning analysis and design can now be substantially automated? If not, where does it expect that expertise to enter the Frontline workflow?

What decisions does Articulate believe should remain with a qualified or experienced learning professional, and how does the product protect those decisions from being overlooked?

My frustration is not based on protecting a job title or reserving software for specialists. Learning professionals have spent years trying to move organisations away from the belief that our role is simply to turn presentations and documents into courses. We have argued for discovery before development, performance analysis before content production and workplace application before information delivery.

Frontline risks reinforcing the very “course factory” mentality many of us have been trying to leave behind, except that the factory can now operate faster and without a learning professional.

The danger is not necessarily that Frontline makes learning expertise unnecessary. It is that it makes that expertise sufficiently invisible for organisations to believe it is unnecessary.

I can see legitimate applications for Frontline. Rapidly changing procedural guidance, product updates, short-lived enablement materials and structured performance support may all benefit from faster creation and maintenance. I am not dismissing the product or its potential.

However, those applications are not evidence that source material can routinely be transformed into effective learning without analysis, professional judgement and quality assurance.

I would genuinely welcome a substantive response from Articulate, not simply reassurance that learning science is embedded in Nova. What has been embedded, how was it validated, where does professional judgement remain necessary, and what safeguards prevent speed of production from being mistaken for quality of learning?

Those answers will determine whether Frontline genuinely extends learning expertise across an organisation or simply makes it easier for that expertise to be ignored.

Articulate seems to be killing Storyline 360 in favour of Frontline where creating the AI slop content is more important than its value and relevance.