Level 1
Foundations
How Claude, ChatGPT, Gemini and the other major assistants actually differ, what each is good at, and how to spot when one is confidently wrong.
From first prompt to first useful result.
Try the lab 75 to 90 min
Private AI lab for senior leaders
Book a session. Leave with the work done.
Seventy-five to ninety minutes, in person or virtual. Your team works on a real problem with the assistant open. Led by Manish Chawda.
Most AI training is a keynote with a demo.
Mine is a lab. It started with four senior executives and twenty minutes each, working on their own tasks with the tools open in front of them. All four became daily users.
4executives in the first lab
20minutes each, hands on
4became daily users
Your tasks, not a demo script
Everyone works on a realistic business problem with the assistant open.
Every major assistant
Claude, ChatGPT, Gemini and others, so you learn where each one is strong and where it fails.
Risk as evidence
Real incidents, from fabricated court citations to deepfake fraud, not hypotheticals.
Three levels. Each one stands alone.
Start where you are. Each level runs 75 to 90 minutes, in person or virtual, from the same material.
Level 2
Real Work
Drafting, analysis, research and decision support on realistic business tasks. Prompting treated as a management skill, not a party trick.
Where AI starts saving you hours.
New courseWrite Better PromptsA 25-minute lab with seven short lessons. Also in a freight and logistics edition.Start the course →Try the lab 75 to 90 min
Level 3
Build & Govern
Agents, automation and the controls around them. What to build, what to buy, and what your board and regulator will ask.
Capability with accountability.
Try the lab 75 to 90 min
How a session runs
Groups of 30 or more, with table helpers so nobody is left behind. Adults learn faster when the room gives them permission to play.
- 1
Warm-up task
A task lands from Kellmoor before the first slide. No theory first.
- 2
Guided lab
You work with the assistant open. I coach from the front; table helpers unstick you.
- 3
Team game
Teams score points on a shared leaderboard for catching AI mistakes and spotting what a human wrote.
- 4
Case file
A real incident, read together: what went wrong, and which control would have caught it.
- 5
Take-away
You leave with the work you started and a habit to repeat on Monday.
One company. Every exercise.
Every lab runs inside Kellmoor Group, a fictional firm with a leadership team you get to know. Tasks arrive from the people who would really send them, so the practice feels like a working Monday.
- Daniel KohManaging Director
- Aisha RahmanCFO
- Marcus LeeHead of People
- Ravi MenonCOO
- Sofia TanHead of Sales
- Helen WardGeneral Counsel
- Jun WeiAnalyst
Games the room plays
AI or human?
Teams judge which text, image or clip was machine-made, then see how often they were fooled.
Spot the AI mistake
A confident answer with an error buried in it. First team to find it scores.
Red-team the bot
Try to talk a company chatbot into something it should never do, then write the rule that stops it.
Live leaderboard
Example scores from a Level 1 game
Table 3460
Table 1390
Table 5320
Table 2230
Case files from the real world
Each incident is read with its sources, so the lesson is evidence rather than anecdote.
- Mata v. Avianca
- Samsung source code leak
- Air Canada chatbot ruling
- Arup deepfake fraud
- Deloitte Australia report
- OpenAI sycophancy rollback
- Chevrolet chatbot
- DPD chatbot
- Microsoft Copilot EchoLeak
- Replit database deletion
- Prompt injection
- Hallucinated citations
- Data leakage
- Deepfake fraud
- Shadow AI
- Over-trusting agents
- Bias in hiring tools
- Unreviewed output
- Vendor lock-in
- Model drift
Built for regulated rooms.
Banks and insurers can add an optional module that maps the session to current Singapore guidance on AI risk. It is taught by someone who has trained senior management at financial institutions on blockchain, risk and cyber.
MAS Guidelines on AI Risk Management
Issued 7 October 2026. Effective 7 October 2027.
We train and advise teams on what it means for their AI. Read the guidelines →
IMDA Model AI Governance Framework for Agentic AI
Launched 22 January 2026, at the World Economic Forum.
We train and advise teams on putting agents under control. Read the announcement →
After the room, keep going.
The classroom starts the journey. A companion app, now in development, carries the practice on with a guided path, practice labs, an AI coach and a prompt library. Companies get a private space for their own team.
- Guided path
- Practice labs
- AI coach
- Prompt library
CoachNIA LabsLed by Manish Chawda.
Founder of NIA and author of the forthcoming The Trust Economy (December 2026). More than 25 years in technology risk, advising banks and building cyber and AI firms from zero. Holds Oxford AI Strategy credentials and has spoken at 40+ industry events.
A practitioner, not a theorist. Every session is run by him, with table helpers.
25+years in technology risk
40+industry events spoken at
3standalone levels
Interactive Session Builder
Configure your room in three clicks.
Pick your leadership cohort, starting point, and format. Your tailored 75 to 90 minute run-of-show updates live and pre-fills the booking brief below.
1Room type
2Starting point
3Format
Bring a lab to your team.
Private cohorts for 30 or more, in person or virtual. Tell me about your team and I will reply personally.
What happens after you sendTwo required fields. One personal reply.
- You send the brief
Only your name and work email are required. The Session Builder can fill in the rest.
- Manish replies personally
A direct email from the person who runs the room. No sales sequence, no mailing list.
- We shape the room together
Your level, your format, and a Kellmoor task that mirrors your team's real work. Then we pick a date.
4 of 4senior executives in the first lab became daily users.