Onshore, Offshore and AI: Who Does What in L&D

Many L&D teams now draw on three sources of capacity: onshore employees, offshore employees, and AI. Used well, the mix lets a team deliver more, faster and at lower cost without giving up quality. Used carelessly, it produces confused handoffs, frustrated clients and content nobody fully owns.

The difference comes down to a simple question asked of every task: who is best placed to do this part of the work?

A practical answer looks like this:

•         Onshore FTEs handle client-facing work. Discovery, relationships, judgment calls and anything where trust and context matter most.

•         Offshore FTEs handle back-end development and clean-up. Building, formatting, QA and the production work that turns a design into a finished product.

•         AI handles first passes, with a developer in the loop. Drafts, outlines, variations and conversions that a skilled person then reviews, corrects and owns.

This post explains why each group fits its role, how work moves between them, and what it takes to make the model run smoothly.

How a project flows between the three

Work starts and ends with the client onshore; AI speeds up the first draft, a designer reviews it, and offshore developers turn the approved design into a finished, tested product.

Onshore FTEs: own the client relationship

Onshore team members share the client's time zone, language, culture and often business context. That makes them the right people for any work where the conversation itself creates the value.

Onshore work typically includes:

•         Discovery and needs analysis. Interviewing stakeholders, digging past the request to the real performance problem, and agreeing what success looks like.

•         Solution design. Choosing the learning approach and making the judgment calls that shape everything downstream.

•         Stakeholder and SME management. Running reviews, negotiating feedback, managing scope and expectations.

•         Live facilitation and presentation. Workshops, pilots, sprint reviews and executive readouts.

•         Final sign-off. Owning the quality of what goes to the client and being accountable for it.

The common thread is judgment in context. These tasks depend on reading a room, understanding organizational politics and building trust over time. They are also the hardest to hand off and the most expensive to get wrong.

Because onshore time is usually the most costly, protect it. Every hour an onshore designer spends reformatting slides or fixing broken links is an hour not spent with the client. The rest of the model exists largely to free onshore people to do this work well.

Offshore FTEs: build, refine and finish

Offshore team members are skilled developers and production specialists. Their strength is turning an approved design into a polished, working product, consistently and at scale.

Offshore work typically includes:

•         Course development. Building e-learning in authoring tools from approved storyboards.

•         Clean-up of AI and draft output. Fixing formatting, aligning to templates and brand standards, tightening structure.

•         Media production. Graphics, video editing, voiceover syncing and animations.

•         Quality assurance. Functional testing, link checks, accessibility checks and LMS packaging.

•         Maintenance. Updating existing courses, refreshing content and handling version control.

This is not "low-value" work. A course with broken navigation, inconsistent styling or accessibility gaps undermines everything the onshore team promised the client. Strong offshore developers often know the authoring tools and standards better than anyone else on the team.

Offshore teams do their best work with clear inputs: an approved storyboard, a style guide, templates and a Definition of Done. Time-zone differences can be an advantage here.

Onshore designers hand off at the end of their day, and finished builds are waiting the next morning.

AI: fast first passes, with a developer in the loop

AI is very good at getting from a blank page to a workable draft. It is not reliable enough to publish on its own. The model that works treats AI as the first step in a task, never the last.

Good uses of AI in the L&D workflow include:

•         First drafts. Course outlines, learning objectives, scripts, scenario ideas and assessment questions.

•         Transformation. Turning a policy document into a job aid, a long course into microlearning, or a script into speaker notes.

•         Variations. Alternative examples for different audiences, regions or roles.

•         Summaries and analysis. Condensing SME interviews, survey comments or pilot feedback.

•         Routine production help. Alt text, captions, translation drafts and quiz formatting.

Developer in the loop means a named person reviews and owns every AI output before it moves forward. That person checks it for accuracy, fit with the design, tone, bias and anything that could expose client or learner data. If something is wrong in the final product, "the AI wrote it" is never an acceptable answer.

The reviewer depends on the output. An onshore designer reviews AI drafts of objectives, scenarios and anything client-specific. An offshore developer reviews AI help with builds, formatting and media. The rule is simple: the person with the right expertise for that output is the one in the loop.

Used this way, AI doesn't replace either team. It shortens the slowest part of most tasks, the start, and frees skilled people to spend their time on judgment and craft.

Quick reference: who owns which task

Task	AI first pass?	Who owns it	Why
Stakeholder discovery and needs analysis	Summarize notes only	Onshore	Trust, context and judgment
Learning objectives and design document	Yes	Onshore	Shapes the whole solution
Storyboards and scripts	Yes	Onshore	Client voice and accuracy
SME and client reviews	No	Onshore	Relationship and negotiation
E-learning build in authoring tool	Partial	Offshore	Tool expertise and consistency
Formatting and template clean-up	Yes	Offshore	Detail-heavy production
Graphics, video and audio editing	Partial	Offshore	Craft and production standards
Assessment question banks	Yes	Onshore drafts, offshore builds	Accuracy first, then production
QA, accessibility and LMS packaging	Partial	Offshore	Repeatable checks
Content updates and maintenance	Yes	Offshore	Volume work against clear rules
Live facilitation and client presentations	No	Onshore	Human presence matters
Final quality sign-off	No	Onshore	Accountability to the client

Treat this as a starting point. Every team's split depends on skills, client contracts and how much context each group holds.

Making the model work

A blended model only pays off when the seams between groups are designed as carefully as the work itself.

Define the handoffs. Agree exactly what each handoff contains: an approved storyboard, asset list, style guide, deadline and named reviewer. Use the same checklist every time. Most delays in blended teams come from incomplete handoffs, not slow work.

Share one set of standards. Templates, brand guidelines, accessibility rules and a written Definition of Done should be identical for everyone. Offshore developers can't meet a standard that lives only in an onshore designer's head.

Use time zones on purpose. Plan handoffs so work moves forward around the clock, and protect a short daily overlap for questions. A single unanswered question can otherwise cost a full day.

Set clear AI rules. Decide which tools are approved, what client or learner data may never be entered, and how AI-assisted content is labeled internally. Check client contracts, since some restrict AI use or offshore access to their materials.

Build one team, not two. Include offshore colleagues in planning, stand-ups and retrospectives. Share client feedback and wins with them. Developers who understand why a course exists make better decisions about how to build it.

Measure the whole flow. Track cycle time, rework and client satisfaction across the full process, not just one group's output. Speed in one stage that creates rework in the next is not a gain.

Pitfalls to avoid

•         Shipping AI output unreviewed. It will eventually contain a factual error, an off-brand phrase or a made-up detail in front of a client. Every output needs a named owner.

•         Treating offshore as a dumping ground. Vague briefs and last-minute handoffs produce poor results, then get blamed on the offshore team. Clear inputs produce clear outputs.

•         Keeping offshore away from context. Developers who never hear client feedback can't anticipate what the client will care about.

•         Using onshore time on production. If senior designers are still fixing formatting, the model isn't working yet.

•         Ignoring contracts and data rules. Confirm what clients allow before their content goes offshore or into an AI tool.

•         Optimizing for cost alone. The goal is the best outcome per dollar, not the lowest cost per hour. Rework and lost client trust erase savings quickly.

The right work in the right hands

A blended L&D team isn't about replacing people with cheaper people or with AI. It's about putting each kind of work where it is done best.

Onshore colleagues build trust and make the judgment calls clients rely on. Offshore colleagues bring production skill, consistency and around-the-clock momentum. AI removes the blank page so both groups start further ahead. A developer in the loop keeps quality and accountability exactly where they belong: with people.

If you're starting out, pick one project and map every task against the three groups. Look for onshore time spent on production, offshore work that waits on unclear handoffs, and first drafts that AI could speed up. Fix those three things, and the model