Glowwellbeing

AI training organised as a curriculum

Glowwellbeing teaches applied artificial intelligence to working professionals in Singapore. Six courses, four capability layers and one shared evaluation standard are used so a team finishes with working artefacts.

See the catalogue How the method works

About

What Glowwellbeing is

Glowwellbeing opened in Singapore in 2026 as a training practice with a single job: turn the use of artificial intelligence inside organisations from one-off experiments into work that can be repeated, checked and handed over. We teach from JTC LaunchPad at one-north and run the same curriculum live for remote participants.

The syllabus covers language models inside existing processes, evaluation of model output, light automation, preparation of source data, and the governance work that sits around those steps. Participants bring a live task from their role. That task is the material of the labs, so the hours spent in class map onto Monday morning.

Classroom

Each module follows the same shape: a short briefing on the mechanism, a worked example on the projector, a laboratory on the learner’s own files, and a review against a written rubric. Tools are treated as replaceable. The method, the coverage matrix and the pass thresholds stay fixed across the 2026 catalogue.

Capability model

Four layers, taught in order

The curriculum is a stack. A layer is closed when the group can perform its tasks against the rubric without a facilitator in the room. Later layers reuse the artefacts from earlier ones, so the same evaluation language appears from Foundations through Capstone.

Literacy

What a model does with a prompt, where it breaks, and how cost, latency and context windows constrain a real task. The layer also fixes a shared vocabulary so a mixed team can talk about output without vague praise.

Practice

Constructing prompts, attaching reference material, and keeping templates that other people can run. Every claim is tested on a small set of cases with a written score, so “it worked once” is not treated as a result.

Systems

Calling a model through an API, chaining steps with checks between them, storing traces, and connecting a sheet, a form or a queue. This layer starts only after Practice has a pass on the rubric.

Governance

Data classes, consent, logging, human review points and a filled risk register for one use case. Governance is taught as practice that sits beside the build, not as a slide deck added at the end.

Levels L1 to L4 sit across those layers. L1 is Literacy under supervision. L2 is independent Practice, with Governance introduced on a single use case. L3 is Systems on a narrow process. L4 is a supervised internal assistant with evaluation set, thresholds and handover pack. We keep Systems closed until Practice is closed because an untested prompt wired into a queue produces volume faster than judgement.

Catalogue

Course catalogue

Hours in the table are contact hours. Laboratory work between blocks is extra and is listed on the full catalogue pages. Formats can be mixed for in-house groups; evening cohorts follow the public calendar.

Course Layer Level Hours Format
AI Foundations for Operating Teams Literacy L1 14 Evening cohort / In-house
Prompt Systems and Output Evaluation Practice L2 21 Evening cohort
Working Data for AI Projects Practice L2 18 In-house
Applied Automation with Model APIs Systems L3 28 Evening cohort
AI Governance and Risk Practice Governance L2 16 In-house / Remote
Capstone: Internal Assistant Build Systems L4 35 Cohort + clinic

Read the level as the independence we expect by the last taught block. L1 work is done with a facilitator in the room. L2 work is repeated on new cases without us. L3 work is a small running system. L4 work is a documented assistant with a named owner inside the organisation. Prerequisites and laboratory hours are on the catalogue page.

Full course pages

Cohort path

How a cohort runs

Evening groups meet twice a week. In-house groups compress the same sequence into taught days and clinics. The five steps below are the spine for both.

  1. 1

    Placement check

    A written task of about twenty minutes. It asks you to brief a model, judge an answer and name one risk in a short process. We use it to place you at L1 or L2, or to recommend a different starting course. It is not an entrance exam with a fail mark.

  2. 2

    Scoping call

    Thirty minutes to choose the working process that will supply lab material: a queue of messages, a report pack, a policy set, a spreadsheet that already drives a team. We agree what can be brought into class and what must stay redacted.

  3. 3

    Taught blocks

    Two or three evenings a week, three hours each, or the in-house equivalent. Each block has a mechanics segment, a guided build and an independent lab. Recordings of the live session are issued to the group the next working day.

  4. 4

    Lab work

    A short assignment on your own files between blocks, typically two to six extra hours per course. You keep a log of attempts. The next block opens with a review of those attempts against the rubric, not against taste.

  5. 5

    Review and handover

    A structured readout: scores on the five criteria, the artefact pack, and a one-page note on what the team can run without us. Capstone groups add a clinic after the last taught week to tidy the prototype and the documentation.

Formats

Delivery formats

The curriculum is the same in every format. What changes is the contact pattern, the group mix and whether the process under study belongs to one employer.

Format Who it suits Schedule Group
Evening cohort Individual professionals Tue+Thu 19:00–22:00, 4–7 weeks 8–16
In-house programme One employer, one process Agreed with the client, usually 2 full days + 3 clinics up to 20
Self-paced with clinics People in shift work Materials on demand + fortnightly live clinic unlimited materials, clinic 12

Teaching is in English. Live remote seats use the same window as the room at one-north; a camera is optional, a working laptop is not. We record taught blocks for people who miss a session. Laboratory time is not recorded because it uses the group’s own files. More on rooms, access and the 2026 calendar is on the delivery page.

Formats and schedule

Artefacts

What participants leave with

Every course produces files the participant can keep using at work. The list below is the shared pack. Course pages name which items appear at which level.

Prompt library for the role

A set of templates with variables, version notes and the cases they were scored on. Colleagues can run them without rewriting the task from memory.

Evaluation sheet with pass thresholds

The five-criterion rubric filled for the live task, with Band 3 marked as the pass line. The sheet is the same format used in class, so later work can be scored the same way.

Mapped process

A one-page map of the chosen workflow with the automated steps marked, the human review points named, and the inputs and outputs written in plain language.

Data-readiness note

What the model can read, which fields were redacted, and which sources are still too messy or too sensitive to use. It is written so an internal owner can act on it.

Risk register for one use case

A filled template covering data, over-reliance, third-party terms and fallback when the model is down. Governance and Capstone both produce a completed copy.

Working prototype

Capstone only. A small internal assistant with a documented evaluation set, logging of runs, and a handover note for whoever will maintain it after the clinic.

FAQ

Questions we are asked most

Do I need a programming background?

Foundations, Prompt Systems, Working Data and Governance are written for people who already run processes, spreadsheets and documents. Applied Automation and Capstone assume you can follow an API example and edit a small script with help in the room. The placement check tells us where to start.

How does placement work?

You complete a twenty-minute written task before the first block. We read the brief, the way you checked an answer, and the risk you named. The result is a recommended starting course, not a pass or fail.

Can this run as an in-house programme?

Yes. An in-house programme uses the same layers and rubric on one employer’s process. Typical shape is two taught days plus three clinics, with the module set chosen after a scoping interview. Group size is up to twenty.

Is there a certificate?

Glowwellbeing issues a certificate of completion listing modules, contact hours and the rubric bands reached. It is a record of this programme. It is not an accredited qualification and not a government certification.

All questions

2026 intake

Request the 2026 syllabus pack

The pack is the current outlines, the coverage matrix, evening dates for April, June, September and November 2026, and a note on in-house scoping. Use the form in the footer, or write through the contact page if you already have a team and a process in mind.

Request a syllabus