Case Study · AI & Video

AI Avatar Employee Onboarding Programme

Multilingual AI presenter-led onboarding across 12 modules. Production time cut by roughly 70% compared with traditional studio video — and every module stays editable long after launch.

Synthesia · Vyond 12 modules Multilingual delivery In production

Produced in Synthesia. No camera, no studio, no reshoots — the presenter, the script and the on-screen graphics all stay editable after launch. Want this for your team?

12Onboarding modules
~70%Less production time
0Studio days required
100%Re-editable after launch
The Brief

Onboarding that scales as fast as hiring does

Most onboarding video ages badly. It is filmed once, in one language, with one presenter — and the moment a policy changes, a tool is replaced or someone leaves the company, the only options are living with stale content or booking the studio again.

This programme was built to remove that trade-off. Every module is presented by an AI avatar driven entirely from a script, which means a content change is a text edit rather than a shoot. New language? Same script, new voice. New policy? Change the paragraph and re-render the affected module alone.

What made it hard

  • Twelve modules covering everything from company story to information security — each needing a consistent presenter, tone and visual system.
  • A multilingual workforce, where subtitles alone were not good enough for day-one material.
  • Content that changes regularly, so the build had to assume revision rather than treat it as an exception.
  • An LMS-ready output: trackable completions, not just video files dropped in a shared drive.

The approach

We scripted the full programme first and locked the instructional design before a single frame was generated — the same discipline as any other course build. Synthesia handles the presenter and voice; Vyond covers the animated explainer sequences where a talking head is the wrong tool. Both feed into one visual system, so the programme reads as a single production rather than twelve separate ones.

The result is an onboarding library a learning team can maintain without ever calling a production crew — and that costs roughly the same to publish in a fifth language as it did in the first.

Inside the Programme

Twelve modules, one continuous journey

Sequenced so a new joiner moves from "who are we" to "here is exactly what you do on Monday" without a gap in between.

Welcome & Company Story
Vision, Mission & Values
How the Organisation Is Structured
Your First Week, Day by Day
Systems & Tools Setup
Code of Conduct
Information Security Essentials
Health, Safety & Wellbeing
HR Policies & Benefits
Performance & Career Framework
How We Work Together
Wrap-Up & Knowledge Check
How It Was Built

Five stages, from script to LMS

The same delivery rhythm every Serve Innovate engagement follows — adapted for an AI-generated production pipeline.

1

Content audit & scripting

Existing handbooks, policies and induction decks were consolidated into twelve scripts written for the ear, not the page — because an avatar reads exactly what you give it.

Module scriptsTone & terminology guide
2

Avatar, voice & visual system

Presenter selection, voice tuning, and a locked template for lower-thirds, callouts and transitions — so every module looks like it came from the same production.

Style frameReusable template
3

Generation & animated inserts

Synthesia renders the presenter segments; Vyond covers process walkthroughs and anything better shown than narrated. Both are cut against the same timeline.

Rendered modulesAnimated sequences
4

Localisation & knowledge checks

Approved scripts are re-voiced per language, and each module closes with a check that reports back to the LMS rather than a passive "next" button.

Language variantsAssessment items
5

Packaging, QA & handover

SCORM/xAPI packaging, cross-device playback testing, and handover of the editable scripts — so the client's own team can revise a module without us.

SCORM / xAPI packageQA reportSource files
The Stack

What each tool is actually for

SynthesiaAI presenter and voice. One script produces the same module in any supported language.
VyondAnimated explainers for processes and scenarios where a talking head adds nothing.
ArticulateKnowledge checks and the SCORM/xAPI wrapper that makes completions trackable.
Your LMSDelivery and reporting. We package to fit the platform you already run.
Common Questions

Before you ask

Do employees actually accept an AI presenter?
For factual, policy-led material, yes — the presenter's job there is clarity and consistency, not personality. We generally recommend keeping the genuinely human moments human: a founder welcome or a team introduction is worth filming properly, and we will say so rather than push everything through the avatar.
How long does a change take after launch?
Editing the script and re-rendering the affected module, not the whole programme — typically same-day for a wording change. The reason is structural: nothing in the pipeline depends on getting a person back in front of a camera.
Which languages can you deliver?
Any language the avatar platform supports, and the list keeps growing. An additional language costs a fraction of the original build because the instructional design, visuals and timing are already done — you are paying for translation and re-voicing, not a second production.
Will it work with our LMS?
We package to SCORM 1.2, SCORM 2004 or xAPI depending on what your platform reads, and test playback on the devices your learners actually use. If you have no LMS, we can deliver a hosted micro-site instead.
Do we own the output?
Yes. On final payment you own the rendered modules, the packages and the source scripts, along with the documentation needed to maintain the programme in-house.

Onboarding you can update in an afternoon

Tell us how many modules you are carrying and which languages you need. We come back within one business day with a realistic scope and timeline.

Share your brief Back to portfolio