Enterprise AI / Field notes from the messy middle
01 / 05I write about what happens when AI meets a real organisation—where economics, identity, implementation and human judgment become more important than the model.
Partner delivery × enterprise AI × operating reality
The thread
Most AI commentary starts with capability. I tend to start with the constraint: the decision that still needs an owner, the evidence that still needs checking, or the workflow that quietly makes the technology irrelevant.
Working models
Compact frameworks for locating the constraint, the cost and the owner around an AI system.
Explore all three →Model 01
Capability starts the conversation. The system around it determines the outcome.
Open the model →Model 02
The cheaper generation becomes, the more value—and cost—moves into verification.
Open the model →Model 03
Proving that an action was authorised does not prove that it was understood.
Open the model →Ideas I keep returning to
These are not finished doctrines. They are positions I keep testing against new systems, markets and operating realities.
01
Generating a plausible answer is becoming cheap. Deciding whether it should change what happens next is where the value—and the liability—now sits.
02
A system can prove who approved an action without proving they understood it. AI agents will make that distinction impossible to ignore.
03
AI has collapsed the marginal cost of interpreting weak signals. Organisations can now hear almost everything; judgment becomes the scarce resource.
04
Models rarely fail alone. Value disappears in ownership, evidence, access, incentives, partner execution and the work required to keep ground truth current.
Selected writing
A build story about why the valuable part of an AI product is often the decision system around the model.
Read on SubstackAI makes software cheap to produce. That shifts the advantage from preserving code to preserving the reasoning behind it.
Read on LinkedInA personal fraud story that exposes the gap between authenticating an action and understanding the intent behind it.
Read on SubstackPlausible AI output is cheap. Enterprise teams pay for the evidence, constraints and maintenance required to act on it.
Read on SubstackExperience
More than a decade across enterprise software, customer value, implementation and partner delivery. The common thread has always been the gap between a strategy and the system required to execute it.
2025—Now
Partner Delivery Assurance Lead, EMEA
Identity made the question of who—or what—is allowed to act inseparable from the question of AI execution.
2024—2025
Partner Services & Delivery Lead, EMEA
Partner-led software only scales when accountability, commercial reality and delivery quality scale with it.
2019—2024
Principal Customer Success Consultant / Sales Solutions Architect
Listening and acting are different organisational capabilities. Better data does not automatically produce a better decision.
2018—2019
Cloud Technology Consultant
The architecture diagram is the easy part. The real work begins when a clean design encounters a brownfield organisation.
2015—2017
SuccessFactors Consultant / Partner Delivery & QA Lead
Implementation quality determines whether strategy survives contact with the people expected to use it.
Bradley FernandesPartner Delivery Assurance Lead, EMEA — Okta
The operator behind the ideas
I work where technology strategy meets delivery reality. That has made me sceptical of clean answers to messy organisational questions.
I publish when I have an argument worth testing—not because the content calendar says it is Tuesday. The writing is mine. AI helps me research, challenge and sharpen it; judgment stays accountable to me.
Follow the next argument ↗