Enterprise AI learning, tracked and access-controlled.
One internal platform that takes every employee from zero to AI practitioner — on our infrastructure, with full visibility, and access issued by our own administrators.
Prepared and presented by Nithin
→ or click to advance · ← back · Esc grid overview · Ctrl+P for PDF
Press N for private presenter notes (opens a second window)
Every team needs AI skills.
Today we have no way to build them.
AI fluency is now a baseline skill
Prompting, model literacy, and safe AI use are expected in every role — most of our workforce has had no structured training.
Per person, per year, external
Commercial course platforms bill an annual licence for every person enrolled — the cost grows with every colleague we add.
Visibility into completion
External training gives us no reliable data on who finished, what they learned, or whether it stuck.
One internal platform. Already built.
The AI Enablement Hub is deployed and running today — not a proposal. It is live at d2525e2y8itcse.cloudfront.net, with Google sign-in working. Three pillars:
Curated public content
The best public learning resources — official docs, talks, and videos — hand-picked and linked (never copied) into one coherent path.
A structured path
An originally-authored curriculum — explanations, deep dives, quizzes, workbooks, and projects — sequenced from zero to practitioner.
Measurable progress
Every hour, quiz score, and completion tracked per employee, inside our environment — with an admin console for L&D.
One core path. Seven tracks.
Everyone takes the same 40 core topics (24.7 h) in plain English. On top of that, each person adds the track that matches their job — five phases, ~17 weeks, a certificate per phase.
Prompt Mastery Track
Everyone — the fastest ROI track for any role
core + track = 30.4 h
12 track topics · 5.8 h · 52 topics total
Engineering Track
Developers, testers, architects, data engineers
core + track = 41.1 h
20 track topics · 16.4 h · 60 topics total
Business & Builder Track
PMs, analysts, admins, marketing, ops, leadership
core + track = 36.5 h
20 track topics · 11.8 h · 60 topics total
AWS AI Track
Engineers, architects and analysts working on AWS
core + track = 32.4 h
10 track topics · 7.8 h · 50 topics total
Azure AI Track
Developers, data engineers and IT in Microsoft shops
core + track = 32.2 h
10 track topics · 7.6 h · 50 topics total
Google Cloud AI Track
Engineers, analysts and ML practitioners on Google Cloud
core + track = 32.3 h
10 track topics · 7.7 h · 50 topics total
NVIDIA AI Track
Platform engineers and anyone self-hosting models
core + track = 32.6 h
10 track topics · 7.9 h · 50 topics total
Foundations
Weeks 1–3
Prompting Mastery
Weeks 4–6
Architectures & Models
Weeks 7–10
Building & Deployment
Weeks 11–14
Ethics, Capstone & Career
Weeks 15–17
Everything you'll see next is running today
It opens with Google single sign-on, working today — the account every employee already has. Microsoft sign-in ships in the same build and switches on from the admin console. No self-signup, no new passwords, and sign-in can be restricted to our own email domains.
Plus spaced-repetition flashcards, notes, bookmarks, comments, and per-phase certificates.
An AI tutor on every topic
- Layered depth — the same question answered from layman ("New to this") to L4 Expert.
- Grounded in our curriculum — answers reference the topic the employee is on.
- Claude-powered — with a built-in offline fallback, so it never goes dark.
- Next: Bedrock + Strands agents running inside our own VPC.
Every employee writes measurably better prompts
A colleague writes a prompt. We score it out of 100 against a published six-dimension rubric — deterministic, computed server-side, not an AI grading an AI — then show them a better one, word by word. 10 guided exercises built on real work scenarios.
Same person, same task, 90 seconds of coaching — then the Lab runs both prompts side by side so they see the difference in the answer, not just in the score.
If it isn't measured, it isn't training
Per employee, automatically
- Topic & phase progress, completions, certificates
- Hours spent learning + video watch time with resume
- Quiz scores across 396 questions
- Prompt Lab scores over time — a real skill curve, per person
- Streaks, weekly goals, and an activity heatmap
Full admin console for L&D
- Manage every topic, quiz, resource, and workbook — no engineers needed
- Per-track analytics — completion and hours split across Prompt Mastery, Engineering and Business
- Sign-in controls — switch Google or Microsoft on and off live, with a guard that can't lock everyone out
- Feature switches for the Prompt Lab and the AI instructor
- Ships today at /admin
Ours to run. Access on our terms.
| External course platforms | AI Enablement Hub | |
|---|---|---|
| Who gets in | Whoever the vendor contract covers | Access keys issued and revoked by our own admins |
| Content | Vendor-owned, changes without us | Our original material, edited by our learning team |
| Infrastructure | The vendor's platform, wherever it runs | Our own AWS account, our own region |
| AI tutor | A separate add-on tier, if offered at all | Built in, on a path to run inside our network |
| Completion data | Vendor dashboards, if any | Full analytics, in our environment |
Nothing about the platform is outside our control: who is enrolled, what they see, and where the data sits are all decisions our administrators make.
Our people's data stays ours
Runs entirely in our environment
Deployed today on our own AWS account — S3 + CloudFront, Lambda and DynamoDB. No learner records, progress data, or usage analytics leave our infrastructure — no third-party learning vendor in the loop.
Clean IP posture
All curriculum text, quizzes, workbooks, and projects are originally authored — ours to keep. Public resources are linked, never copied; videos play through YouTube's official embed API.
- Google / Microsoft SSO only — no self-signup, so no orphan accounts
- Optional email-domain allowlist; local passwords exist only for named admins
- AI tutor roadmap keeps inference in our VPC (Bedrock)
- Degrades gracefully offline — no external service is required to run
Ready for the whole company today
The build is finished and deployed. Google sign-in works. Access keys decide who gets in, issued from the admin console. There is no engineering work left between this room and a rollout — only a decision about who and how fast.
Built and deployed
- 132 topics across 7 tracks, fully tracked
- On AWS: S3 + CloudFront, Lambda + DynamoDB
- Google sign-in working — open it right now
- Prompt Lab + AI instructor with fallback
Opened by invitation
- People sign in with the work Google account they already have, then enter their access key
- Access keys control entry — issue, revoke, reuse, from the admin console
- Microsoft sign-in ships in the same build and switches on live
Team by team, or all at once
- Stage it with access keys: one team, then the next
- Or issue keys org-wide on day one — technically identical
- Hours, scores and activity are visible from the first day either way
Early concepts — not commitments
This platform is finished. These are the next things I am exploring. They are ideas at this stage, nothing more — I would rather show you the direction than surprise you with it later.
AIBuzz
An AI-focused product concept. Still taking shape — I am working out what it is and who it is for.
PAN
A product idea for performance testing. Early thinking, nothing built yet.
Published research
An intent to write up the approach behind this platform and publish it as a paper.
Shared as a signal, not a proposal — so you can see this is a pipeline, not a one-off.
Access-controlled, rolled out deliberately. Three doors — your call.
The platform is built, deployed and running on our own AWS account. Access is not open — people request an access key and an administrator issues it, so you decide exactly who gets in and when. That gives you three doors.
Use it
Run it exactly as it is today. Administrators issue access keys, team by team or org-wide, and you keep control of who is in at every step.
One decision · admin-controlled rolloutSupport it
Put it on a company AWS account and give L&D the admin console, so key issuing, revocation and reporting sit with the org rather than with me.
Optional · org owns the admin consoleOwn it
If leadership sees strategic value, the company can take this on as its own product — including offering it externally as an add-on service. That is a conversation worth having.
Terms open to negotiateWhat I would like back is simple: real employees using it. Live usage is the one thing I cannot build myself — it is what tells us what to fix, sharpen and add next.
Live right now. Access by key.
Yours if you want it.
Deployed, signed in and working — 132 topics, 89.6 h of curriculum, seven tracks, with access keys issued by our own administrators. Use it, support it, or own it: the door is yours to pick.
Open the live platform →