Team Topologies as the Infrastructure for Agency — Matthew Skelton

Matthew Skelton

Open Menu<br>Close Menu

Team Topologies as the Infrastructure for Agency

11 May

Written By Matthew Skelton - Holistic Innovation

“Organizations that succeed with AI are not necessarily those with the most advanced models. They are the ones that: Start with clear outcomes; Organize around value flow; Bound agency thoughtfully; And actively share what works”

— Matthew Skelton

Summary: effective knowledge work at speed - whether with humans or with AI - requires ‘bounded agency’ to be effective

Photo by Anderson Wei from Pexels https://www.pexels.com/photo/modern-elevated-highway-with-concrete-pillars-33473266/

Why AI progress is stalling<br>It feels like many organisations are still searching for tangible returns from AI.<br>There’s no shortage of investment or experimentation. But outcomes are often uneven.<br>At QCon London, I suggested that many of these challenges are not technical. They’re organisational. That aligns with what we’re seeing more broadly: despite the focus on models and tooling, the real constraint is often how teams are structured and how decisions are made.<br>This is increasingly reflected in wider research. For example, Where AI Delivers Value Today, found that organisations seeing value from AI tend to redesign workflows and decision-making, not just introduce new tools.<br>Bounded agency as the missing piece<br>A useful way to understand this is through bounded agency .<br>Who (or what) is allowed to act, and under what conditions?<br>As I’ve mentioned in my keynote, trust in both teams and AI systems comes from having clear boundaries- defined missions, explicit guardrails, and a well-understood scope of action. I explain in my blog that this holds whether the work is done by humans or AI: clarity of context and constraints leads to better outcomes. Without this, things tend to drift.<br>Some organisations are effectively doing the opposite by granting AI tools broad, sometimes unbounded access to data and systems.<br>It’s worth pausing on that.<br>We would rarely give a human team unrestricted access across the organisation. Yet with AI, this often happens by default.<br>The risk isn’t just security. It’s a loss of clarity about ownership, purpose, and how value is created.<br>Cognitive load, for humans and AI<br>There’s a parallel here that’s easy to overlook.<br>For humans, too much context increases cognitive load and slows decision-making. For AI systems, the same pattern appears: exceed the context window, and coherence starts to break down.<br>In both cases, bounded scope improves outcomes.<br>This aligns with findings from Gartner, which consistently highlight that AI initiatives deliver stronger ROI when organisations define clear domains, responsibilities, and governance structures, rather than expanding access indiscriminately.<br>Team Topologies as Infrastructure<br>This is where Team Topologies becomes useful.<br>The patterns and principles - clear team boundaries, independently viable services, and well-defined interaction modes - act as an infrastructure for agency .<br>They create the conditions where both humans and AI can operate effectively:<br>Clear boundaries protect domain fidelity

Cognitive load stays within manageable limits

Capabilities are exposed through explicit interfaces

Teams can act with autonomy while remaining aligned

In many ways, this isn’t new. Organisations that have already focused on these principles for human teams are often better positioned to adopt AI effectively.

Discover how Team Topologies acts as the Infrastructure for Agency in humans and AI - keynote

Why knowledge diffusion matters<br>Structure alone isn’t enough.<br>A common challenge is that progress stays localised through small pockets of success that don’t spread.<br>How do good practices move across the organisation?<br>This is where Adapt Together acts as a useful complement. It focuses on the active diffusion of knowledge and innovation by creating alignment, building trust, and enabling multiple independent flows of value to evolve over time.<br>We’ve seen this approach lead to meaningful outcomes: reduced dependencies, faster delivery, and measurable cost efficiencies at scale.

Discover how Adapt Together™ from Conflux drives effective alignment and innovation diffusion at scale

A quieter shift in leadership focus<br>A pattern seems to be forming.<br>Organisations that succeed with AI are not necessarily those with the most advanced models.<br>They are the ones that:<br>Start with clear outcomes

Organise around value flow

Bound agency thoughtfully

And actively share what works

The underlying principles are not new.<br>But the context has changed.<br>AI is making the gaps in organisational design more visible and more urgent to address.

Request a keynote talk

AITeam TopologiesAdapt TogetherLeadershipConflux

Matthew Skelton - Holistic Innovation

Co-author of the award-winning and ground-breaking book Team Topologies, Founder and CEO/CTO at Conflux, and director of...