Weeknotes 480

Roger on 10/10/2025

I did:

Starting

Started lots of new things this week.

  • Started a new business case for some future work.
  • Started meeting people from a new team I’m joining soon.
  • Thought of some ideas for an upcoming AI hackathon.
  • Watched a presentation on some brilliant service design for how to improve outcomes for students furthest from advantage.
  • Started planning a workshop about opportunity space mapping.
  • Wrote a list of 4000 profanities for blocking from our platform. The things I get up to.
  • Chatted about playbooks and how they should be the “collective wisdom of multiple voices”, not an instruction manual for the one and only way to do things.
  • Set my annual objectives. I’m not usually keen on personal objectives when everything we do is in collaboration with others, but I’m interested in setting a baseline and seeing how much I can achieve in twelve months.

I read:

Autonomy and alignment

I’ve been trying to understand autonomy and alignment better so I watched and read:

And learned:

  • Autonomous agile teams increase employee motivation and job satisfaction significantly as well as boost creativity and productivity.
  • The biggest barriers to autonomy are teams understanding overall direction (goals are often set by management without involving the teams, that they are often equal to deliverables and deadlines so team members don’t know what the goals are) and managing external dependencies (which requires extra, often distracting work).
  • Formal (e.g., standardised reporting) and informal (e.g., shared values) control mechanisms are necessary for organisational alignment.
  • Enabling constraints (e.g., work in progress limits) help teams have autonomy.
  • Feeling a heightened sense of responsibility is an important cultural element for achieving the balance between autonomy and alignment.

My conclusion is there is a big difference between autonomy and alignment and autonomy in alignment. If you ensure alignment first through enforced control mechanisms and only give teams autonomy within those controls, then you aren’t balancing autonomy and alignment.

Releasing the work at the right rate

A roadmap for successful AI adoption in Higher Education

I respectfully disagree. Not only does the article not mention any of the challenges facing higher education institutions, it talks about starting with well defined problems, but then only problems that can be solved by AI.

I think it’s important to remember that AI and its economic benefit is still just a big bet, it isn’t proven yet. I’m leaning towards it having a revolutionary impact on society but that taking decades to pan out, so in the short-term here’s my roadmap for success with AI (in higher education or any sector):

  • Understand your users. If AI bursts, you’ve got a better understanding of who uses your products and services.
  • Fix your data. If AI doesn’t pan out, you’ve got your data in good shape.
  • Improve your content. If AI fails, you’ve got great content.
  • Train your people. If AI doesn’t replace everyone, you’ve got really capable people.
  • Improve your operational processes. If AI doesn’t automate everything, you’ve got a more efficient organisation.
  • Encourage a culture of innovation. If AI doesn’t live up to its promise, you’re able to respond to the next big change.
  • Coach your teams. If AI doesn’t become the ultimate assistant, you’ve got great teams that can tackle complex problems.
  • Educate your leaders. If AI doesn’t create new business models, your leaders will.
  • Update your technology. If AI isn’t the next big tech trend, you’ll have reliable, secure systems.
  • Run experiments. Even if all your experiments with AI fails, you’ll know how to learn from evidence.

Sort all those things out and AI is easy. Don’t sort them out and nothing you do with AI will succeed.

Thinking about managing

Started reading about the history of management for the first module of my MBA. Interesting insight: cross-functional teams in coal mines were studied back in 1951 for the efficiency gains they had over functional teams with hand-offs.

I thought:

What we get wrong about prioritisation

In his talk about team trust, Tom Dolan mentioned the challenge of prioritisation is (and I’m paraphrasing) that we’re not comparing apples with apples, we’re comparing apples with oranges with a pear tree with an orchard with the concept of horticulture. There is no way to make a rational comparison. So the answer is to not compare. The answer is asking the right questions to decide whether to go forward with the opportunity. Every opportunity should be assessed on it’s own merits.

Future of work

Adam Smith gave us the phrase ‘division of labour’ in 1776 and the factory owners of the industrial revolution turned it into the standard concept for organising work around specialists who perform one part of all the work required to create something of value. Since then pretty much every job has been based on that idea, whether it’s manufacturing, service or knowledge work, each person only knows how to do their little bit.

AI might be the thing that finally changes centuries of industrial thinking about how to organise work. Rather than people with limited specialist knowledge moving the work onto the next person once they’ve done their bit, we’ll have people who are domain generalists doing the human stuff like relationships and navigating messy social systems working with AI that has the specialist knowledge in the domain.

Health care is an obvious example of a domain that could see this change. A patient wouldn’t see a nurse for triage, a radiologist for an x-ray and a doctor for treatment. In this new world, they’d see one person who would use AI to do the triage, x-ray and treatment.

Privilege distance

If power distance matters within organisations, then privilege distance matters between organisations and users.

Get weeknotes in your inbox