From Geek to Star #43 - Enabling AI in your organisation as a Tech Leader (2/3)

If you missed the previous episodes, you can access them online here.

🗓️ This Week – Episode 44: AI-driven engineering as a transformational change

In the previous newsletter, I shared my first thoughts on what, at a company level, can “Enabling with AI” mean when there is so much potential power together with so many risks related to unleashing AI in an organisation.  I shared how AI enablement for an organisation revolves around 3 dimensions and touched upon the first one: 

1. AI augmentation of people and teams across functions: how to augment people across all functions, not just tech, thanks to AI. Not only at an individual level but also at a team collaboration level. 

2. AI-driven software engineering: how to fundamentally change software engineering practices with the usage of AI, up to potentially revising the sourcing strategy of the company between internal and outsourcing capabilities. 

3. AI-augmented cybersecurity: how to integrate AI in the cybersecurity operating model of the company in face of the exponential threats brought by AI on the other side, but also to manage the acceleration of digital assets in the company due to dimensions 1/ and 2/ above.

In this newsletter, I will share on Dimension 2: AI-driven engineering. For this, I will share my learnings from an excellent long and detailed article that Sau Sheong, CTO of GovTech Singapore wrote on approaching AI-driven engineering at scale called “The Context and The Harness”. 

Insights and learnings from Sau Sheong Chang: from “prompting” to engineering systems

Reading Sau Sheong’s article “The Context and the Harness” was a bit of a “pause and think” moment for me. Because it puts words on something many of us are starting to feel but are having to structure properly:  AI-driven engineering is not about writing better prompts. It is about designing systems around AI.

And I believe this is where many organisations - especially traditional ones - are still underestimating the shift.

Insight #1: Prompting is the new “hello world”

For the past 18 months, a lot of focus has been on:

• prompt engineering

• choosing the right model

• benchmarking tools

Useful but insufficient at scale with many teams involved. As Sau Sheong explains, when AI evolves from a chatbot to an agent interacting with systems, tools and workflows, the prompt becomes just one component of a much larger environment .

This environment includes:

• system instructions

• company knowledge

• APIs and tools

• memory

• workflows

In other words, we are no longer prompting a model, we are engineering its environment. 

Insight #2: Context is the real leverage

One idea I found particularly striking is this: AI systems don’t “know”. They predict based on what they know. Which leads to a very practical consequence:

The quality of output is directly driven by the quality of context we provide.

This shifts the role of engineers from writing logic to curating, structuring and controlling context

Concrete examples:

• ensuring documentation is usable by AI (not just humans) - something that historically engineers have always considered doing reluctantly

• structuring knowledge so retrieval is meaningful - something that companies are usually not good at, as they consider knowledge management as a waste of time.

• defining clear system instructions and constraints

• designing examples that guide behaviour

This also explains something many of us have experienced: same model, same prompt style but very different results across organisations

Because the difference is not mostly the model, it is the context.

Insight #3: You need a “harness” to trust AI

This is probably the most important shift for tech leaders. Even with perfect context, AI remains:

• probabilistic

• non-deterministic

• sometimes wrong (confidently wrong)

Sau Sheong introduces the concept of a “harness”: Everything you build around AI to make it reliable, testable and safe to use.

Think about it:

• tests

• validation

• monitoring

• guardrails

• fallback mechanisms

This is not new in software engineering. But what is new is applying it to something that is not deterministic by nature.

Which means:

• we don’t test exact outputs

• we test acceptable ranges of behaviour

• we don’t just deploy

• we continuously evaluate and adjust

AI systems are not “build and run” as we are used to, they are “design, measure, refine  continuously”

Insight #4: The real work is not where people think

One of the most interesting takeaways from the article: The hardest part of AI-driven engineering is not the AI. It is:

• the platform quality

• the documentation

• the standards

• the engineering discipline

Because AI will amplify whatever is already there. If:

• documentation is messy → AI produces messy outputs

• standards are inconsistent → AI creates inconsistencies at scale

• architecture is unclear → AI introduces chaos faster

AI doesn’t fix our engineering maturity. It exposes it and amplifies it.

Insight #5: From coding to system design

Putting all of this together, what emerges is a clear shift. The role of engineers is evolving from writing code to:

• designing systems that produce code

• defining constraints

• orchestrating workflows

• ensuring reliability

What this means as a tech leader or engineer

If I simplify everything into a few actionable thoughts:

• Stop thinking tools first

• Start thinking system design around AI

• Invest in:

  • documentation

  • standards

  • knowledge structuring

• Build:

  • evaluation loops

  • guardrails

  • observability

• Accept that:

  • AI will not be perfect

  • but we can design systems where it is reliable enough

🙏 I’d Love to Hear From You

How advanced do you feel your team is today in terms of maturity about AI-driven engineering?

Reply to this email, I read every note.

Follow me on LinkedIn for more reflections and “behind-the-scenes” thinking between newsletters. Don’t hesitate to comment or reshare,  it’s one of the best ways to grow your SHINE 🌟. If you want to know more about how I can support you or your teams to thrive in a tech career in this AI-age, have a look at my offerings here.  

P.S. Referral Pilot 🚀

Forward this email to one engineer or tech friend who could also benefit from this newsletter: sharing is caring - a little gesture can go a long way to strengthen bonds.

✨ May the SHINE be with you!

From Geek to Star by Khang | The Way Forward