Writings
Thoughts on AI engineering, flow metrics, building products, and lessons from 25+ years in tech.
That’s not good.
Before, I was orchestrating up to ten concurrent coding agents. Now I’m orchestrating up to ten concurrent orchestrators—each running its own fleet of up to ten concurrent sub-agents.
The second signal is in: lead time in minutes and hours, not days and months.
Measured on a production SaaS codebase (n=153, intent → main): - P50 (median): 26m - Avg: 1h 22m - P85: 2h 43m - P98: 10h 10m Against the Value Stream Flow Report (2023) industry baseline, this is an order-of-magnitude shift. It's becoming clearer: agentic coding is an operating-model shift, not a plug-in.
I kept hearing “AI coding = marginal gains.” My experience didn’t match that at all, so I measured.
TL;DR: one product-engineer orchestrating a fleet of agents reached ~22× more net new code and ~2× better net/churn efficiency in a real product codebases (not toy project). Curious how this lands with engineering leaders — does this stack up against your numbers?
Dunbar's Number and Operating Model Design👇🏽
Robin Dunbar, Emeritus Professor of Evolutionary Psychology at Oxford, studied how social bonds form in primates and humans. His work helps us understand the limits of social relationships and their impact on team cohesion, trust, and structure.
🧠 Cognitive Load and Shifting Complexity in Domains 👇🏽
Cognitive load changes as the complexity of domains shifts over time. What is simple today may become complicated or complex tomorrow.
🧠 Cognitive Load And Its Elements 👇🏽
The Cognitive Load Theory, which explains how the human brain processes and manages information, provides fascinating insights into why processing information feels harder or easier depending on where we are in our journey. It discusses 'ELEMENTS'—individual units of information that our brain processes—in our working memory and how these change as we gain expertise.
Cognitive Load vs Domains 👇🏽
Teams designed to be self-sufficient often can take on too many responsibilities, leading to a high 𝗰𝗼𝗴𝗻𝗶𝘁𝗶𝘃𝗲 𝗹𝗼𝗮𝗱. Cognitive load is directly influenced by the number of 𝗱𝗼𝗺𝗮𝗶𝗻𝘀 a team is responsible for.
The Different Types of Cognitive Load 👇🏽
Did you know that fellow Australian John Sweller, Professor Emeritus at the University of New South Wales, developed the Cognitive Load Theory? How cool!
🍽️ Dinner conversation: Passion vs Grit 👇🏽
The advice often given to young people—' just follow your passion'—doesn't sit well with me. I have two young humans trying to figure out this world and how to navigate it and reconcile their feelings with their thoughts and all the noise and nonsense on social media.
Action is the only thing that turns dreams into reality.
Last week, someone told me they couldn't share their next GenAI big idea—it was too precious and too secret. I laughed because I recognised my young self at that moment.
Agentic Future: Routines and Handoffs 👇🏽
I've been exploring autonomous agents for a while now, working on a series of different use cases. The power they'll bring in creating smart workflows is mind-blowing.
Kanban Australia 2024
The conference is just around the corner, featuring a fantastic lineup of speakers! Here's a quick look at what's in store: ➡ Keynote Speaker Patrick Steyaert will discuss how ambiguity challenges organisations balancing discovery and delivery and how value can be created in this transitional space.
Ever feel held back by your own thoughts?
Mindtraps are habitual patterns of thinking. They distort our perceptions and lead to unhelpful beliefs and behaviours.
Where Should Digital and Tech Teams Call Home?
The dilemma of where digital and tech teams should establish their home base is one that shapes an organisation's ability to innovate and deliver value effectively. Should these teams be embedded directly into business units for maximum alignment and speed?
The ways teams choose to interact can make or break the effectiveness of any operating model.
In the context of operating model design, teams can be organised around 𝗮𝘀𝘀𝗲𝘁𝘀, 𝗲𝘅𝗽𝗲𝗿𝘁𝗶𝘀𝗲, 𝗼𝘂𝘁𝗰𝗼𝗺𝗲𝘀, 𝗼𝗿, 𝗴𝗲𝗼𝗴𝗿𝗮𝗽𝗵𝘆. 💡 Choosing the optimal 'organising principle' for your teams is crucial: ➡ 𝗢𝘂𝘁𝗰𝗼𝗺𝗲𝘀-𝗕𝗮𝘀𝗲𝗱: Teams organised to achieve specific outcomes, such as Event, Mission, and Project Teams.
Why is Autonomy Such a Powerful Magnet for Teams?
This week, I'm kicking off a new advisory work with one of the ASX Top 100 companies through Neu21, focusing on designing an effective operating model that empowers teams through autonomy. It got me thinking: why is autonomy such a powerful magnet?
We spend so much time and energy trying to become better individuals, parents, professionals, and leaders.
I've missed gatherings, postponed catch-ups, neglected replying to messages, and let "busy" become an excuse, assuming my friends would understand because we're all navigating our own challenges. Lately, I've found myself becoming more introspective.
Flow Awareness = Flow Efficiency 🌀
Resource Efficiency and Flow Efficiency are two different operations strategies. 👇🏽 ▶️ One focuses on maximising the utilisation of individual resources, and the other focuses on maximising the value delivered to customers.
Flow Awareness = Waste Awareness ♻️
How much waste is hidden in your delivery ecosystem? Days?
Flow Awareness = Better Culture 💪
Another 5 unmistakable improvements consistently observed in Flow-Aware organisations: 👇🏽 1️⃣ Increased Trust Flow Awareness significantly increases on-time delivery. Flow-aware teams deliver more, more frequently, sooner, and at higher quality.
Flow Awareness = Competitive Edge
5 Unmistakably Clear, Tangible Improvements Consistently Observed in Flow-Aware Organisations: 👇 1️⃣ ➖ Significant Reduction of Lead Time ➖ Reducing lead time means reducing the time to learn. With new ideas in users’ hands sooner, organisations can learn and evolve faster.
🧩 Why Isn't Flow a More Common Topic Among Executive Leadership?
Flow remains an underutilised concept in many enterprise organisations today, and I wonder why. 🤔 In my experience, organisational change towards embracing flow usually starts with a visionary and proactive leader—someone who supports the early seeds of flow and champions it among their peers.
🌀 The Ambiguity of Prioritisation.
I see so much ambiguity everywhere I go! This confusion is not just annoying.
I started looking at #Flow from an observability perspective.
As in, Flow Observability. 👇 The term observability is originally borrowed from control theory, where the “observability” of a system measures how well its internal state can be determined from its outputs.
❗ The Problem with *Mono-Mode* Operating Model Strategies
There’s a lot of hype around building autonomous cross-functional teams, and while the idea is great, it doesn’t fully fit within the complexity and depth of enterprise organisations. 👇 A mono-mode operating model that focuses solely on autonomy isn’t enough.
In many large organisations, product teams are also service teams.
👇 The Product + Service Paradigm 👇 There’s a significant paradigm shift that many enterprise product teams haven’t fully embraced yet—the dual role of being both product builders and service providers. These teams aren’t just building products; they’re also providing ongoing services to internal or external stakeholders by maintaining, enhancing, and supporting the products they create.
The Myth of [Fully] Autonomous Teams in Enterprise Organisations
I'll be upfront about it: there’s no such thing as fully unbounded autonomy in enterprise organisations. 👇 We keep hearing about the push for Autonomous Teams in large enterprises, especially driven by the big consultancies who seem to pitch the same PowerPoint deck everywhere (remember when they were “spotifying” everyone?
👇 Here are 10 different ways to CONSTRAIN DEMAND in large-scale digital portfolios:
Building on yesterday’s post, balancing capacity and demand is the secret to getting a system into a state of flow. Again, there are two levers: Increase Capacity and Constrain Demand.
The secret to getting a system into a state of flow is balancing capacity and demand.
Two levers: Increase Capacity and Constrain Demand. There’s a misconception that increasing capacity means increasing headcount.
Here are different ways to think about capacity in your delivery ecosystem:
👇 1️⃣ Individual Capacity Measured in hours, skills, and mental space available to a person. This fluctuates daily and varies individual by individual.