Choose “Save as PDF” in the print window. Landscape, no margins, background graphics on.

Operator at work on industrial equipment

DevLearn 2026 · Expo Stage

Building a digital brain for manufacturing

Connect your tools, capture knowledge from real work, and prove who can do the job

Thomas Bellamy · Co-founder and CRO, Edplay

Who can run line 3 tonight?

Count how many systems you’d open to answer that.

2 / 15

The problem

Your plant has the answer. It’s in six places.

HR system

Who works here, on which shift

LMS

Who completed which course

Competency matrix

Who was signed off, and when

Quality system

Deviations, NCRs, CAPAs, audit findings

Maintenance

What broke, and who fixed it

The expert’s head

How the job is really done

3 / 15

The digital brain

Combine what you already run into one capability map.

Workforce and shifts

Competency records

Quality and audit data

Operational data

Real work, captured

→

Workforce intelligence

Every person, skill, machine and site in one view. Approved knowledge linked to each task.

→

OUT

Skill gaps to train

Knowledge to capture

Answers at the machine

Proof of competence

4 / 15

The key decision

Does this team need training, or does the knowledge need capturing?

Skill gap

A team is at 61% against an 85% standard. The approved training exists.

→ Assign it to that group.

Knowledge gap

There isn’t enough approved material to teach it. It lives in someone’s head.

→ Capture the real work first.

“Most quality defects arise from procedure deviations, incomplete understanding of requirements and inconsistent execution.” · Senior operations leader, medical device manufacturing

5 / 15

When the gap is knowledge

Capture it, verify it, train on it.

Line clearance on a contact lens blister packaging line, captured hands-free and turned into an approved course.

6 / 15

At the point of need · Ask Lia

Put the approved answer right at the machine.

Operator: Line 2 filler, station 4. Seal check keeps failing after changeover.

Lia: Here’s the approved post-changeover seal check for this machine, with the video of how it was fixed last time. Isolation steps first.

Source: approved SOP and recorded repair on this equipment

Describe the problem and location

Get the approved step mid-job, in your language.

Every repair tied to its machine

The next person sees what was found and fixed last time.

Questions feed the map

What people ask shows where the gaps really are.

Approved SOPs, task videos and linked skills, in the worker's language.

7 / 15

Approved training · At the point of work

Real work becomes training people can use.

Keep the real context

Build training and onboarding from existing SOPs, task videos and the expert's explanation.

Short and visual

Use interactive formats and microlearning, aimed at the learning objective and audience.

Keep the approval gate

AI prepares the draft. People review and publish the method the floor will follow.

One standard template for every SOP and course.

8 / 15

The models are interchangeable · The evidence stays

Keep the source, sign-off and retest together.

Machine-linked knowledge

Keep the approved procedure, task recording and reviewer sign-off against the task and equipment.

The outcome record

Know which training addressed which gap, on which task, and how many retests it took.

A connected loop

Gap detection becomes useful when it leads to verified training, retesting and recorded competence.

AI content alone is not proof that someone can do the job.

9 / 15

For L&D in the room

Speak operations. These numbers get investment approved.

Financial

Sustained yield improvement
Labour efficiency
Material cost reduction

Operational

OEE
Downtime and changeover
First-pass yield
Units per labour hour

Quality and compliance

Audit findings
NCRs and CAPAs
Safety incidents
Customer complaints

10 / 15

How the return is measured

Your numbers before, then month by month.

Gap closure rate. Gaps closed against your baseline, on your own competence scale.

Time to competence. Days from a gap being found to qualified sign-off.

Retest pass rate. Share passing on the first retest, not just completion.

Reviewer edit rate. How much qualified reviewers change the AI draft.

Weeks until a new starter works alone · −58%

BeforeMonth 3Month 6Month 12

Minutes to fix a repeat fault · −67%

BeforeMonth 3Month 6Month 12

Compare against your own baseline, not industry averages. Connect it to scrap, downtime or first-pass yield.

11 / 15

Connect and secure

Connect the systems you already run.

Learning managementHR systemsQuality and auditMaintenance and work orders

Built to connect with: SAP · Workday · ADP · Paylocity · ServiceNow · Oracle

Your environment or ours. On-premise, your cloud tenant, or isolated single-tenant.

Zero external model training. Your SOPs and recordings never train outside AI.

Encrypted end to end. AES-256 at rest, TLS 1.3 in transit.

Governed by your rules. Bystander face protection, role-based access to footage.

12 / 15

How to start

Three phases, one platform.

Phase 1 · Foundation

Create and standardise

Turn the SOPs, videos and training you already have into approved material, in one standard template.

Moves: hours per item, time to independent work

Phase 2 · Insight

Connect and see

Connect the systems you already run, see who knows what on which machine, and find the gaps.

Moves: skill gaps closed, unplanned stops

Phase 3 · Scale

Capture and roll out

Add hands-free capture on the floor and roll out across lines and sites.

Moves: time to fix a fault, scrap and rework

The 12-week pilotWeeks 1 to 4: Connect and baselineWeeks 5 to 8: Capture and verifyWeeks 9 to 12: Train and prove

12 work areas225 contributors32 to 4 headsets4Core connectors set up

Proposed scope, not a commitment. Leave with trusted knowledge, measured evidence and a rollout decision.

13 / 15

Take back to your plant

Three things to try on Monday.

Ask “who can run line 3 tonight?” Count the systems it takes to answer.

Sort your open gaps: skill gap or knowledge gap?

Tie your next project to one number operations already tracks.

14 / 15

Get in touch

Visit us at Booth 631

Viktor Potapov, CEO

Cell:(512) 687-4417

Email:victor@edplay.ai

Thomas Bellamy, CRO

Cell:(512) 659-6177

Email:thomas@edplay.ai

Connect with us

Viktor Potapov

Thomas Bellamy

Book a call

Scan QR code

Thank you