I recently returned from the LeSS Conference in Berlin and the Global Scrum Gathering in Amsterdam. Along the way I participated in training on Beyond Budgeting (with Bjarte Bogsnes) and Designing Agile Organisations (with Cesario Ramos).
After not travelling to a conference since early 2020, it was a fantastic learning experience featuring many of the world’s leading thinkers and Agile adoption experiences. This month’s feature article is inspired by a big theme at both conferences: how AI will impact us.
As with any prediction, keep in mind what Neils Bohr said:
“it’s difficult to make predictions, especially about the future”.
Goodbye all team members being human, hello AI team
A lot of us are using ChatGPT and the like for idea generation and learning. The “Agile AI Guy”. We’re now seeing such AI Learning Facilitators (AILF’s) trained specifically on Agile concepts.
Snehal Talati is known as the “Agile AI Guy”. I attended his session at the 2023 Global Scrum Gathering in Amsterdam at which he showed a preview of his AgileGPT product. He showed examples of it generating a User Story with Acceptance criteria for Single Sign-On (SSO) feature. Then creating advice for a technical architect on what to consider in a SSO implementation. There’s no doubt that it can be a time saver. Also no doubt that you should not turn your brain off and just copy-paste without customising to your context.
An increasing number of programmers are using Github’s Copilot to generate code, document their code, create tests etc. It certainly is impressive. Personally, I turned of Copilot whilst learning Python as I felt that it was preventing me from thinking everything through myself.
Are you interested having the benefits of pair programming but can’t convince other human team members to pair with you? Cursor is the first IDE designed from the ground up for pair programming with an AI. It’s always ready and willing to pair with you!
Snehal sees the future Agile team as comprising not only human members but AI agents as well. Be warned: you may see stakeholders sending AI bots to your Daily Scrums to bring back a summary!
Goodbye narrow specialisation, hello multi-learning
A real revelation is how much specialised AI’s can already produce at least basic versions of what narrow specialists do but orders of magnitude quicker and cheaper.
This includes tasks traditionally done by product managers, business analysts, UI designers, testers, frontend programmers, database designers and solution architects.
In his keynote at the LeSS conference, Craig Larman introduced us to Generative-ai and LLM-Assisted Development (GLAD) and projected the following time savings for various tasks.
This has huge implications for the composition of our Agile teams. Why would we need specialist business analysts, UI designers, testers, frontend programmers, database designers and solution architects when AI allows someone operating much more broadly to do all of that in far less time?
The full keynote is now available online.
Goodbye Scrum secretaries, hello capability leaders
Spinach.io is an example of an AI tool that takes meeting notes, captures action items and proposes Jira issues. In the Atlassian marketplace it is promoted as “your AI Scrum Master”.
Of course, Spinach.io is not covering anything close to the Scrum Master role. Unfortunately there are many people called Scrum Master today whose time is dominated by similarly administrative activities to this. Booking meetings, updating the progress tracking tool, reporting etc. This is little different than the activities that happened in traditional projects by a “Project Co-ordinator”. This anti-pattern has be called “Scrum Secretary”.
The good news is that tools like Spinach.io and the myriad of other AI Learning Facilitators (AILF’s) are set to take over the administrative drudgery that have held Scrum Secretaries back from spending their time in impactful ways. This includes pursuing non-trivial organisational impediments, uplifting the product ownership capability, teaching managers about how to take a systemic approach to organisational improvement etc. These are all part of the real Scrum Master role and tools and skills to pursue them are taught as part of the Advanced Certified ScrumMaster (A-CSM) and Certified Scrum Professional-ScrumMasster (CSP-SM) courses accredited by Scrum Alliance.
Question for WOW leaders
“Is the direction that we are going in with our ways of working aligned with our AI disrupted future?”
Here are some implications of this that I expect will be controversial.
- If you are continuing to increase the number of specialisations, then the answer is no.
- If you are sticking with single function line management or “chapters” inspired by the so-called “Spotify model”, then the answer is no.
- If you are designing your teams around owning components in the technical architecture (“component teams”) or minimising cognitive load (as per Team Topologies), the answer is no.
- If you are adding roles focused on co-ordination as part of your approach to scaling up, then the answer is no.
- If there is no effort to move your Scrum Masters beyond being tied up with co-ordination and/or administrative tasks, then the answer is no.
It seems inevitable that AI will in a small number of years be capable enough to do many tasks done by narrow specialists as well or better in a small fraction of the time and cost. For private sector companies, it then becomes a matter of remaining competitive to adopt such technology and redesign teams around this. Even for public sector organisations I expect that it will eventually look irresponsible to spend an order of magnitude or more than necessary on such work.
We’ll still need humans for creative problem solving, empathising with customers, teaching and coaching people to do all of those better and better. Multi-skilled product developers are likely to work at a higher level of abstraction than today e.g. orchestrating micro-services. If already you’re into one or more of those paths, you have no need to worry about your employability. For others, there’s a little more adapting to do.
