Case study access

This case study is password protected.

Enter the password to continue.

← Back to work Case Study 06 · Design Thinking · UX · Visual Design

AI Powered Training Management System

A unified platform that turns fragmented training operations into a guided journey, from course discovery and approval, to enrollment and progress tracking.

ClientAventis Learning Group
RoleDesign Thinking, UX & Visual Designer
Cloud partnerIBM Cloud
DurationDiscovery to MVP
Aventis recommended courses, sector and skill filters with AI match percentages
The problem

Course admin was still running on spreadsheets and email.

Aventis has been managing its courses, scheduling and coordination with clients and training providers through traditional means, e-mails, spreadsheets and phone calls. As Aventis' business and client base grew, there was a strong urgency to improve the efficiency of course management and communications, so Aventis could keep its commitments to end-users and keep expanding.

HR administrators were struggling to cope with the workload, since current training administration was very manual and time-consuming. Aventis wanted a training management platform that let HR officers manage training and learning for employees org-wide, while also helping individuals enhance their own skills.

The platform needed to:

  • Manage, track and monitor course enrolment.
  • Give personalised course recommendations based on AI.

Technology decision point, Data Fabric.

IBM Cloud CDN Email Delivery Service Cloud Image Registry Cloud Object Storage Cloud Function Cloudant Code Engine Container Registry Watson NLU
My role

Three hats across one engagement.

Design thinking facilitator
Facilitated the design thinking workshop that aligned stakeholders around a business-centric outcome and validated the problem statement.

UX & interaction designer
Built interactive prototypes and wireframes, defined use cases end to end, and ran usability studies, planning and gathering feedback through user interviews and online surveys.

Visual designer
Translated validated requirements into style guides, graphic assets, icons, design patterns and pixel-perfect interfaces.

The process

From business framing to a built MVP, in six stages.

01

Business Framing

Confirmed the executive sponsor and product owner, and determined which domain use cases to focus on, initiative exploration, vision definition, opportunity canvas and opportunity statement.

02

Technical Discovery

Understood the user's as-is journey and top pain points, and explored technical areas of interest and non-functional requirements, security, integration, edge and more.

03

Design Thinking

Identified a series of MVPs, defined the hypothesis to be tested, and surfaced risks, assumptions and UX/UI requirements through big ideas, prioritisation and a to-be scenario.

04

Inception

Defined a secure minimum viable architecture that mitigates risk, crafted user stories, refined MVP scope, and validated dependencies through user flows and prototyping.

05

MVP Build

Tested the hypothesis via a learning-driven production release against the stated business outcome, and developed the UI components enabling specific user requirements.

06

Build out

Iterated across multiple MVPs, refining the solution until the stated business outcome was achieved.

Transformation & results

Two sessions reframed the entire brief.

Design thinking workshop. A workshop I facilitated clarified the requirements, the target users, and the scope of the MVP as the outcome IBM and Aventis agreed on. It left us with a clearer picture of the solution to build and the priorities to focus on.

Data discovery session. Building an AI-driven platform meant first exploring Aventis' existing data, surfacing the limitations and possibilities for the recommendation engine. Most of it turned out to be unstructured and unlabelled, so automated labelling became a necessary part of the IBM solution.

Implementation

Co-created and co-executed in three agile sprints.

IBM and Aventis co-created and co-executed the MVP in an agile scrum manner, proving the AI-Powered Training Management System could:

Register & view course details Search for relevant courses Automated course labelling Personalised recommendations View popular courses Manage budgets Automated enrolment emails

Beyond the benefits to users, Aventis also gained a standardised, automated data collection pipeline.

User research

Meet Sharon, HR Manager.

40, with a wealth of experience as a human resource manager. She holds a bachelor's degree and is in charge of training and development for the whole company.

"I struggle to cope with the workload as current training administration is very manual and time-consuming. How can I book course(s) for my employees to overcome these challenges and the workload?"

She needs the course-finding and booking process to be faster and more effective, and support to reduce her workload and streamline her day-to-day schedule.

Too many tabs open to compare courses
No visibility into approvals
Budget checks done manually
Recommendations based on guesswork
Follow-ups lost in email
No single source of truth
As-is scenario

Seven steps, every one of them manual.

Sourcing, finalising, registering with Aventis, informing employees, preparation, tracking and post-training activity, every step ran through email threads and spreadsheets, with Sharon coordinating manually across trainers, management approval and her own team.

Sourcing Finalising Registration Informing the employees Preparation Tracking Post training
Pain points

Pain points, reframed as opportunities.

Top pain points

  • Finding the right course and provider, based on budget and staff availability.
  • Collating and preparing reports was a painstaking process.
  • Co-ordinating and registering for a course was a painful process.
  • Tracking progress and feedback.
  • Employees struggled to justify or convince their employer.

How might we…

  • How might we find the right course based on staff availability and requirements?
  • How might we make the reporting process dynamic, with everything in one place?
  • How might we digitise the co-ordination and registration process?
Big ideas

Three concepts carried into the MVP.

Magic Pot (AI)
An AI solution to find the right course based on topics, availability, budgets and competency frameworks.

User journey map
A virtual map of the training courses an employee has taken, mapped against job scope, a skills radar and KPIs.

Dashboard view
An interactive, easy-to-access dashboard to review past trainings, budgets, enrolled courses and invoices.

To-be scenario

From a Google search to a booking, guided end to end.

Sharon lands on the Aventis website looking for a training provider. The Magic Pot AI recommendation helps her find a course suited to her requirements, and she registers and books her interest.

She receives an email with a course summary, the trainer's profile and background, and a link to book and pay, which she forwards to her management for approval.

Once approved, she books and pays for the course. Her Aventis profile now has a dashboard showing payment status, budget utilised, course timelines and training status.

Discover Search Registration Booking Preparation Payment Dashboard
Vision & hypothesis

A 50% reduction in time-to-book.

"A human resource leader will have an improved admin experience to reduce the steps and time taken to find and book the right course by 50%."

We assume that…

  • HR is aware of her team's competency levels.
  • The Magic Pot is accurate and able to recommend relevant courses.
  • We have a good set of data available for our use.
  • Employees are aware of their own competency skills and level.
  • Companies are able to share their data with IBM/Aventis.
  • The quality of courses and trainers on TMS is comparable to offerings elsewhere.

If this is not true, then…

  • She'll pick an unsuitable course and her team may struggle.
  • No one will trust or use the tool.
  • The data will need to be cleansed, adding time to build.
  • We won't be able to map the now-and-future skills picture.
  • We won't have the correct, polished data to work with.
  • People will shop elsewhere.
Prioritising risks

Ranked by how known, and how costly, each risk was.

We prioritised assumptions first by how known or unknown they were to us at the time, and second by how much impact each assumption and risk would have on the success of the solution, plotting them across a known/unknown, high/low-impact matrix to decide what to validate first.

MVP statement

What Sharon's website needed to do.

01
Find and book a course relevant to her employees
02
See a personalised dashboard of progress, budgets and enrolled courses
03
See an intuitive map of the course journey
04
Easily share and notify employees with relevant information
Next case study

Digitised PPE Compliance Checklist & Real-time AI Dashboard →