Anyone can ship an AI product this afternoon. Almost no one will.

Kyle Balmer’s AI with Kyle Daily Update 188 dropped this week with a claim that should make every L&D director sit up: you can build, ship and sell a real AI product in an afternoon. Not a prototype. A real one. With a landing page, a login, a payment button, and an output a stranger would pay for.

He’s right. I rebuilt a small internal tool with Claude Code last week in about the time it takes to make a cup of tea. Twenty months ago that would have needed a small dev team and a fortnight.

The reason this matters for everyone reading this isn’t whether you personally are going to spin up a side hustle this Saturday. The reason it matters is that the people in your business who are paid to think, plan, communicate, sell, recruit, finance, market or coach are all suddenly working in a world where the barrier between “I had an idea” and “I built it” has collapsed.

That collapse changes what good looks like in almost every department. And it’s the thing our AI apprenticeships are built to teach.

Kyle’s funnel: the unglamorous truth about AI advantage

Kyle made a point that’s worth quoting:

This is the bit that gets misunderstood about AI in 2026. The advantage doesn’t go to the company with the smartest people. It goes to the company with the most people who actually do the thing.

The advantage goes to the team where the HR Business Partner has actually built a Copilot Studio agent for recruitment screening. Where the Finance Manager has actually automated the month-end pull. Where the Marketing Coordinator has actually built a Zapier flow that pulls leads into the CRM. Where the Service Engineer has actually used AI to triage a callout.

Not the team where they read about it. Not the team where leadership talked about it at the SLT meeting. The team where someone shipped.

Specific. Not clever.

Kyle’s second great line was on specificity. The mistake people make when they think about AI in their team is they reach for “AI-powered business assistant.” Pants, as Kyle would say. The thing that actually works is: “turn messy product notes into a five-email launch sequence.”

Specific. One input. One output. One painful job. One clear result.

That’s the same shape we drill into every learner on our AI & Automation Practitioner Level 4 apprenticeship. Module 3 (Spotting Waste & Inefficiency) is literally about hunting your own organisation for these one-job opportunities. Module 5 (No-Code Automation Tools) is about building them in Power Automate, Zapier, Make or Copilot Studio. Module 11 closes with AI agent design in Copilot Studio and Gemini Gems.

It’s not theory. By the end of month three, our apprentices have automated something on their actual floor that saves their actual team actual time.

Why apprenticeships, not a weekend YouTube binge?

Fair challenge. Why would anyone need a 15-month apprenticeship for skills that, on the surface, you could learn on a Saturday with Kyle’s prompt and a Claude subscription?

Three reasons.

First, scale. One person who reads Kyle’s email and builds a thing is great. Forty people across your business all building things, with shared governance, shared prompts, shared playbooks and shared assessment, is transformational. That’s what a closed cohort gives you that a weekend hack does not.

Second, the bit Kyle didn’t mention. Governance, GDPR, AI risk, the EU AI Act, ISO 42001, audit trails, vendor risk on AI tools, ethical adoption, change management at scale. The reason most enterprises don’t ship is not the code. It’s the procurement and risk wrap around the code. Our standard covers all of that, formally assessed. Our UK AI compliance guide covers the regulatory layer in depth.

Third, it’s funded. Levy-payers pay nothing. Non-levy employers pay 5% of an £18,000 programme (about to shift to 25%, worth acting before the cliff). Standalone AI Leadership Units are 100% government-funded for SMEs and £750 per learner for levy-payers. £0 to £750 to put someone through structured AI capability development that ends with a Microsoft credential and (depending on route) a CMI Level 3, Level 5 or BCS Foundation in AI? That maths works.

What this looks like inside a real business

One of our learners last quarter, a Sales Manager at a logistics client, used Module 7 (Data-Driven Decision Making) to run an analysis on 18 months of pipeline data. They found their deals were stalling at the same stage. They restructured the process. Win rates jumped 15%.

That’s a Kyle-style story. Specific input (pipeline data). Specific process (Copilot analysis). Specific output (restructured process). Specific result (15% improvement). Inside a real job, paid for by the levy, with their line manager backing them.

That’s what the funnel looks like with structure around it. We’re not selling a different course. We’re selling the structure that gets people from the top of Kyle’s funnel down to the tiny puddle of people who actually do the thing.

The TESS suite, mapped to Kyle’s principles

If the case for AI capability has just clicked for you and you’re looking at your team thinking “so who, in what programme, when”, here’s the quick map:

The honest summary

Kyle’s right. You can ship a real AI product this afternoon. The hard part is no longer the code. It’s the discipline, the specificity, the governance, the scale and the willingness to actually go and do the thing.

That’s what we teach. That’s what gets funded. That’s what your team needs.

If this resonates and you’ve got people in your business who could be the ones in the tiny puddle that Kyle talks about — the people who actually ship — drop me a line. Fifteen minutes, no commitment, just a chat about who in your team is in the right place to start.

Frequently Asked Questions

About the Author

Related Reading