By Nathan de la Haye
15/06/2026

Last month, I attended AWS Summit London 2026 at ExCeL London. I attended several useful sessions across AI agents, workforce change, innovation culture and future-ready systems.

Below, I have attempted to distil the main learnings into some takeaways for myself and the wider Digital Jersey community.

I hope the reflections below are helpful, and I’d be very happy to discuss any of them further.

TL;DR

The main message I took from AWS Summit London was that AI is continuing to move from individual tools into end-to-end business workflows.

The first phase of generative AI adoption has largely been about people using AI to draft text, summarise documents, generate ideas, write code or speed up everyday tasks.

Now the focus is shifting towards AI-enabled workflows, agentic systems, new team structures and more practical questions around governance, skills and adoption.

In Jersey, our economy is built around service-heavy, knowledge-intensive and regulated sectors, which are exactly the areas where AI could create value.

But they are also the areas where adoption needs to be particularly controlled and properly governed.

The age of agents is really about workflows

A central theme at AWS Summit was “The Age of Agents”.

Agentic AI refers to systems that can do more than respond to a single prompt. An AI agent can be designed to take steps through a workflow, use tools, retrieve information, complete tasks and support decisions.

For most Jersey organisations, the immediate opportunity is likely to be AI supporting repeatable tasks with staff required to review and stay accountable.

That could mean:

Helping a compliance team search internal policies.
Supporting a fund administration team with document summaries.
Helping a public sector team triage requests.
Assisting a customer service team with first-draft responses.
Supporting developers with code generation, testing and documentation.

Future-ready AI systems need governance from the start

Another session I found useful focused on how to design future-ready agentic systems.

An AI tool used to draft a low-risk internal document is one thing. An AI system that interacts with client data, supports compliance activity, handles operational tasks or influences decision-making is another.

In regulated sectors, a key concern is whether AI processes can be governed, evidenced and explained.

Future-ready systems need clear data boundaries, permission controls, audit logs, testing, monitoring, escalation routes and human review.

The starting point should be bounded use cases: internal knowledge retrieval, policy mapping, operational checklists, meeting-note summaries, document triage or first-draft analysis.

These are areas where AI can support productivity while keeping responsibility with trained staff.

This aligns closely with the AI Playbook for Jersey, which focuses on responsible and effective AI adoption across the Island’s economy.

A curious workforce needs clear guardrails

Several sessions at AWS Summit focused on innovation culture and the importance of building a curious workforce.

In many organisations, the people closest to the work are those best placed to identify useful AI opportunities. They know which processes are slow, where documents are hard to navigate, where handovers create delays, and where customers or colleagues repeatedly ask the same questions.

But unmanaged experimentation creates risk, especially where staff are using public tools, sensitive data or unapproved workflows.

The challenge for Jersey businesses is to create safe ways to experiment, giving teams permission to test AI, but within clear rules.

Team structures will need to adapt

Another useful theme was team structures in an agentic world.

As AI moves from individual tools into workflows, responsibility for adoption also needs to shift.

A successful use case usually needs several perspectives: the business team that owns the process, the technology team that understands systems and data, the risk or compliance team that understands controls, and the operational team that understands how the work actually gets done.

Team structures in an agentic world will require cross-functional teams around clear AI implementation use cases.

Skills need to become more role-specific

The workforce discussion also reinforced that AI capability is not a single skill.

A board member does not need the same AI knowledge as a developer.

A compliance officer, relationship manager, civil servant and operations lead will each need to understand different risks and opportunities.

In Jersey, we know that advanced digital skills remain a challenge, and that we need to move more quickly from discussion to delivery.

This is the reason that our Digital Jersey AI training is intended to be more practical than simple awareness.

Teams need to learn how to identify useful workflows, write better prompts, check outputs, protect data, assess vendors, escalate risks and measure whether AI is actually improving the work.

The takeaway for me is that our AI programme needs to continue to be tied to real roles and real processes.

The Jersey opportunity

The main message I took from AWS Summit London was that AI is now entering a more operational phase and is being embedded into workflows and teams.

The most useful approach to agentic AI is to start with a few practical steps.

Identify repeatable workflows with manual, document-heavy or information-heavy tasks.
Assess the risk and determine where human review is essential.
Run controlled pilots with approved tools.
Build skills around real roles and real workflows, rather than generic AI awareness.

As a small, connected jurisdiction, Jersey should be efficient at bringing businesses, technologists, policymakers, educators and regulators into the same conversation more easily than larger markets.

Digital Jersey has a useful role to play here: sharing guidance, convening sectors, supporting skills development, connecting businesses with expertise and helping organisations move from interest to safe experimentation.