Financial services
Enterprise agentic AI
Target architecture, a federated control plane and an agentic AI evaluation framework for a Tier-1 global bank.
Enterprise agentic AI London, UK
I’m Adam - I help enterprises take AI from pilot to production, with measurable business impact. From boardroom strategy to hands-on engineering, I design and build the systems and teams that make it happen.
01 / From intent to executionAJ—26
From the first decision to runtime.+
Experience across
Capgemini Invent · PwC · Contino · HPE
£20m
Acquisition where I was instrumental in delivering,
through technical and proposition strategy.
25+
People in the R&D team I led.
Customer time-to-value reduced by 45%.
Board to build.
Executive advisory, architecture
and hands-on AI engineering.
01 / Selected builds
Ideas, made operational.
Evaluation & reliability01
01
Workflow02
Evidence03
EvaluationTrace → test → evidence↗
CAESAR
Making agentic workflows measurable. An evaluation platform combining generated tests, trace analysis and explicit safety gates.
Applied researchAgent orchestration02
01
Plan02
Build + review03
VerifyHuman direction. Coordinated agents.↗
Metis
The meta-harness that orchestrates AI coding agents - deterministic dispatch, scope as a contract, cross-vendor review, enforced by exit codes instead of hope.
In use by developersAIMux
A service mesh and control plane for multi-agent systems.
Tested in the wild02 / Enterprise practice
Built around
real-world complexity.
My work at Capgemini Invent spans financial services, government, telecommunications and energy. The common thread: making AI useful, governable and operable.
Public sector
AI security at government scale
Reference architectures and design patterns incorporated into cross-government AI guidance.
Energy & utilities
A national AI agent marketplace
End-to-end platform architecture for the governance, orchestration and execution of hundreds of autonomous agents.
03 / How I work
One conversation.
The whole picture.
Commercial context, architectural judgment and engineering detail belong in the same room.
Strategy & advisory
Shape the opportunity
Turn an ambiguous AI ambition into a clear investment case, operating model and delivery roadmap. Technical judgment for executives and boards.
Architecture & governance
Design the system
Connect agent platforms, evaluation, identity and policy into an architecture that fits the realities of a regulated enterprise.
Engineering & leadership
Spearhead the delivery
Set the technical direction, build alongside engineers and lead multidisciplinary teams from a scoped problem to a working system.
04 / Field notes
Thinking in public.
Machine Learning Experiments Need Observability
KerasTuner handles search, but experiment review needs a lightweight observability layer around tuning.
The Allure of the MVP
Why your AI prototype and your production system aren't on the same trajectory.
Governance That Accelerates Delivery
Governance should reduce uncertainty and increase deployment speed, not block progress. Here's how to make that real.
A useful starting point