Inside the book
The argument, in short.
What stops AI programs in regulated institutions, what putting humans in the lead actually means in practice, and the frameworks the book hands you to act on it.
← Back to the bookThe biggest challenges around AI in financial services aren’t technical. They’re human.
Trust. Leadership. The people who build AI, the people who use it, the people it affects, and the people who are supposed to govern it. Get those things right and the technology works brilliantly. Get them wrong and you end up with expensive experiments, regulatory headaches, and an organisation more confused than when you started.
- —
Governance gaps. Nobody clearly owns the AI’s outcomes. Business, technology, risk and compliance each assume someone else is responsible.
- —
Change resistance. Employees fear for their jobs or simply don’t trust the outputs, so they work around the system rather than with it.
- —
Customer friction. AI experiences are designed to reduce cost rather than serve customers — and the difference shows.
- —
Regulatory blind spots. Teams build first and worry about compliance later, only to find that “later” means months of rework.
- —
Measurement myopia. Success is defined in efficiency terms — cost, speed, headcount — while trust, adoption and fairness go unmeasured.
Every one of these is a people problem, not a technology problem. Yet most AI strategies treat people as an afterthought. That’s backward, and it’s why so many efforts fail.
What “Human in the Lead” means
It doesn’t mean a human approves every AI decision — that’s impractical and often unnecessary. It means humans set the intent, hold the accountability and own the consequences. AI absorbs the low-value work, surfaces the right information, and makes people faster and better informed. People do the things only people can do: build relationships, exercise judgment, show empathy, navigate complexity.
- 01
AI should augment people, not just automate tasks. The goal isn’t to remove humans from the equation. It’s to make humans better at what matters most.
- 02
Accountability must be clear and shared. When an AI system makes a decision that affects a customer or the business, a named human owns the outcome. Not “the algorithm.” Not “the data science team.”
- 03
Trust is built through transparency, not just performance. A highly accurate system can still destroy trust if people can’t see how it works or challenge its decisions.
- 04
Employees are partners in change, not victims of it. The organisations that win invest in their people rather than surprising them with automation.
- 05
Governance is an enabler, not an obstacle. Done well, it gives you the confidence to move faster and scale further.
Named frameworks, not principles.
Every chapter hands you something you can put in front of a steering committee on Monday.
Three Models of Human Involvement
Human-in-the-Loop, Human-in-the-Command, Human-in-the-Lead, with a decision framework for matching each to a use case.
The Attention Budget
Why oversight that looks rigorous on paper becomes a rubber stamp, and how to design intervention points that don’t.
The Role Redesign Canvas
For rebuilding jobs rather than automating tasks.
The Oversight Design Canvas
And a reviewer training curriculum outline.
The AI Governance Maturity Model
Plus an AI system classification template.
The 90-Day Roadmap
A day-by-day plan for launching or resetting an AI initiative.
Foreword by Sanjay Kukreja, Chief Technology Officer, eClerx
“What distinguishes Human in the Lead from the shelf of AI books written for executives is that it refuses the false choice between speed and safety. The institutions that thrive will not be the ones that adopted it first, but the ones that adopted it in a way their clients and employees could believe in.”
Get the first chapters before the book ships.
- —Read chapter previews as they’re released
- —See the frameworks applied to live case vignettes
- —Invitations to roundtables and reader sessions for executive teams
Subscribers get the template library ahead of publication — the AI system classification template, the Oversight Design Canvas and the Role Redesign Canvas, in editable form.