How to reduce course development time from 190 hours to under a day

The figure that shocks most people is not the 190 hours itself. It's what they discover when they break it down - that the majority of that time is not being spent on the thing that creates value.

The research benchmark of 190 hours to develop one hour of interactive e-learning has been cited in L&D for over a decade. The Chapman Alliance study (2010) that produced it is frequently referenced, occasionally contested, but broadly consistent with what L&D professionals report when asked how long their courses actually take.

The more important number is what that time is for. Because when you look at where the 190 hours go, it becomes clear that the problem isn't the work - it's which work is taking the most time.

Where 190 hours actually goes

For a standard 60-minute interactive e-learning course built with a traditional authoring tool - Articulate Storyline or Rise, iSpring, Lectora - the time breaks down roughly like this:

• Needs analysis and brief development: 10-15 hours. Stakeholder conversations, defining objectives, scoping content, agreeing assessment criteria.
• Content research and SME extraction: 20-30 hours. Sourcing subject matter expertise, reviewing existing materials, conducting SME interviews, synthesising into a course outline.
• Script and storyboard writing: 30-40 hours. Writing every word of every section, designing the interaction and assessment flow, getting stakeholder sign-off on content before build begins.
• Build in authoring tool: 50-70 hours. Creating slides or pages, configuring interactions, building quiz logic, adding media, applying branding.
• Review and revision cycles: 20-30 hours. Stakeholder review, SME review, quality assurance, feedback incorporation, often two or three rounds.
• Testing and publishing: 5-10 hours. LMS testing, SCORM packaging, learner acceptance testing, final sign-off.

Total: somewhere between 135 and 195 hours depending on complexity, organisation size and number of stakeholders involved.

"The majority of those hours are not spent on instructional design. They are spent on production. And production is the part that AI can handle."

The hidden time costs

The figures above assume a relatively smooth process. In practice, three things add significant hidden time that rarely appears in project estimates:

Waiting time. The 190 hours is working time. Total elapsed time is typically three to four times longer, because courses wait - for SME availability, for stakeholder review slots, for sign-off from people who are busy doing other things. A course that takes 190 hours to build routinely takes 12 to 16 weeks to finish.

Rework caused by late feedback. When a stakeholder reviews a fully built course and requests structural changes - not just copy edits, but changes to the flow, the scope or the assessment design - the cost of that change in a built authoring file is dramatically higher than it would have been at the scripting stage. Late-stage rework is one of the most consistent sources of budget overrun in e-learning projects.

Version management. Compliance courses in particular require regular updates as regulations change. With traditional tools, each update is effectively a new project - open the source file, make changes, republish, re-upload to the LMS, re-communicate to learners. The maintenance burden of a content library built on traditional tools grows with every course added to it.

What AI changes - and what it doesn't

The promise of AI in course creation is that it compresses the production stages dramatically. The reality is more nuanced than the marketing suggests - and understanding the nuance matters for setting realistic expectations.

What AI changes:

• Content research and synthesis - AI can read source materials and extract the relevant information into a structured course outline in minutes rather than days.
• Script writing - AI generates a complete first draft of every section, including scenario content and assessment questions, removing the blank-page problem entirely.
• Build time - in platforms where AI generates the course directly into the authoring environment (rather than producing a script you then have to build), the build stage collapses from 50-70 hours to near zero.
• Update cycles - AI that can compare an existing course against updated source material and identify what's changed removes the need to rebuild from scratch every time a policy updates.

What AI doesn't change:

• Needs analysis. Understanding what the course is actually trying to achieve, for whom, in what context, still requires human judgement and stakeholder conversation.
• Expert review. A subject matter expert still needs to verify that the generated content is accurate. AI can produce plausible content about almost any subject - but plausible is not the same as correct, and in compliance or regulated contexts, the difference is significant.
• Final quality judgement. An experienced L&D professional reviewing the output and applying professional standards before publication still adds value. The goal is not to remove human expertise from the process - it's to focus it on the stages where it matters most.

The 7-hour breakdown

When the production stages are compressed by AI, the time breakdown changes fundamentally. Here is what a comparable course looks like when built with CourseAgent:

• Brief and objectives: 30-45 minutes. Write or upload the course brief. Define the audience, objectives and tone. AI uses this to structure the entire generation.
• Source material upload: 15-30 minutes. Upload any existing documents, policies, PDFs or videos. AI reads and incorporates them into the course structure automatically.
• AI generation and review: 1-2 hours. AI generates the full course - structure, content, scenarios, assessments. Review the output against your objectives. Most users make targeted edits rather than wholesale rewrites.
• Quality audit: 30 minutes. Run the 6-layer AI quality audit. Review any flagged items and address them.
• Stakeholder review: 1-2 hours. Because the course already exists in a reviewable form, stakeholder review is faster and feedback is more actionable. Reviewers are commenting on a real course rather than a storyboard document.
• Final edits and publishing: 30-60 minutes. Incorporate feedback, final checks, publish or export SCORM.

Total: 4 to 7 hours, depending on complexity and the number of stakeholders involved.

Time comparison: traditional vs AI-assisted

• Content research and scripting: 50-70 hrs → 1-2 hrs
• Build in authoring tool: 50-70 hrs → 0 hrs (AI generates directly)
• Review and revision: 20-30 hrs → 1-2 hrs (reviewing, not rebuilding)
• Testing and publishing: 5-10 hrs → 30-60 mins
• Total: 190 hrs → 4-7 hrs

The quality question

The obvious challenge to all of this is: does speed come at the cost of quality?

The honest answer is: it depends entirely on what you mean by quality, and on which AI tool you're using.

If quality means visual polish - complex animations, bespoke interactions, pixel-perfect branding - then yes, an AI-generated course will not match what a specialist designer can produce in Storyline with 60 hours of build time. That trade-off is real.

But if quality means instructional effectiveness - whether learners actually acquire the knowledge and capability the course was designed to build - then an AI course that has been generated with solid instructional design principles can match or exceed a traditionally built course, because the limiting factor in traditional production is rarely the designer's skill. It's the time available to apply it.

When a designer has 190 hours, they can apply rigorous instructional design. When they have 190 hours divided across five simultaneous projects, they can't. AI gives that time back.

Where to start

If you're looking to reduce course development time in your organisation or consultancy, the most practical starting point is a single project comparison.

Take a course you need to build - ideally a compliance or knowledge-based course rather than a complex simulation - and build it twice: once with your current process, and once with an AI authoring tool. Measure the time at each stage. Compare the output quality using a structured rubric.

The result of that comparison will tell you more than any benchmark figure - because it will be specific to your context, your standards and your learners.

Try it with CourseAgent

CourseAgent's free plan lets you build up to three complete courses - with no credit card and no time limit. The average time from sign-up to publish-ready course is under 30 minutes.

Start your first course free →